==============================================================================
===  TO DO  ==================================================================
==============================================================================
{{{ Other people's bugs
    Reduce -nobuild: wrong placement for het O3 H wrt R141 and D215 in 2DRI
    Changes still made by Reduce -nobuild:
        Existing protonation state for His
        Existing Met CH3 and Lys NH3 orientation (don't want -norotnh3 and -norotmet)
    Reduce: protonates S in Cys--Fe bond (1jb0)
    Reduce: H not added to end of MSE in std PDB files [Daniel Farrell]
    Probe: Weird intra-sidechain clashes for 1BKR (low occ atoms in-chain)
    Probe: clashes in 1F7L het because Cs named (sort of) like Ps
        Clashes in Coenzyme A ligands (A/P atom names, eg 1bo4)
    NOEdisplay: NOEs thicker, esp. violations
    Prekin: doesn't mark pointmasters on the start of vectors (those flagged P)
    Prekin: dangling altB C--N for 1SWU B 29 (whenever i has all A and B alts)
    Prekin: Prekin/Win keeps kin file open continuously
    Prekin: stray vectors on windows -- buffer flush issue?
    IE6: doesn't display transparent PNGs correctly (gray background)
    IE6: hitting enter submits form without a cmd=[button name] (worked around)
}}}
{{{ Ideas, long-term goals, etc.
    User manual (do we need this in addition to page-by-page help?)
        Tutorial / walk-thru / Captivate movies?
        Write about maintaining MolProbity
        Write about features
    Could make it a web service with XML-RPC (SOAP, CORBA too complex) or REST
    Create better uptime management tools, including memory usage (and jobs?)
    Estimate time for e.g. analysis jobs?
    Add "priviledged" users who can e.g. run extra NMR models
    Allow uploads of tarballs of NMR models?
    On upload of alt conf model, do table to choose which ones to use
    "nice" large jobs such as multi-model NMR runs
    Allow simple comparison of bad residues in two models; see jobs/aacgeom.php
    Add documentation to the analysis results page (how to interpret)
    Make sure Reduce doesn't shrink the PDB file (lost atoms/models, etc)
    Add coloring type drop box for -lots kins
    Support "permevents" for notebook pages - eval() or just a pageCall() target
        Wanted this so e.g. analysis results notebook could link to Reduce or SSWING
    analyze::findAltConfs() should also support nuc. acid mc/sc classification
    Allow escape from job when no process number appears after x sec.
    Add whois query to session listing
    Change upload_setup's fetch of EDS map to be like welcome.php (a la carte)
    Develop "helper" pages; mark optional steps clearly
    DANG is included in bin/ but we don't *use* it anywhere... / upgrade to 1.9
    Move labbook.html to charts/ ?
    Auto-uniquify chain IDs on file upload?
    Bob: add % favored/allowed directly to kinemage graphics?
    SwingSet: auto-select res that are outliers (store in SESSION/modelID/__?__)
    Use ELBOW to add H to unknown hets
        Alternately: For unrecognized hets, determine connectivity using Molikin,
        write HET dict, allow user to remove H using HTML textbox, save to server.
    Produce multi-model Ramachandran plots from the GUI
    Tool for building ensembles from arbitrary sets of models
    Convert "number of H added, etc" text into tables
    Auto-expand coords, kins folders?
    add real HTML head/foot to downloadable multichart? look into viewer issues
    Can upload_xxx_done pages be converted to work with generic_done?
    Add documentation at the bottom of all pages:
        Add H
        Edit PDB files
        Make simple kinemages
        Interface contacts
        Analyze
        View and download files
    Mitch Miller: how should Rama stats respond to alternate conformations?
        Don't over-count alts; maybe weight by occupancy?
    add "ini_set('memory_limit', -1)" to scripts in cmdline to remove mem limit?
    convert USER  MOD to REMARK USER MOD
}}}
    Add multichart to format (CSV) that would work with KiNG, Excel, etc.
        Residue by residue list of outliers (y/n) -- CSV?  standalone? [PHENIX]
    Convert to biol. unit should produce and display a notebook entry
    Reduce doesn't recognize HSD (as HIS) or ILE _CD_
    we don't correctly auto-detect CNS format nucleic acid PDBs on upload
    Change "Go back" buttons to "Cancel" buttons?
    Note that NMR models should be done one at a time to get multichart, etc
    Add page to explain meaning of MolProbity score
    record file contents: protein, nuc. acid, or both (?)
    write improved to-do list for Coot w/ clustered errors (waiting on Paul)
        merge this clustering func. with analyze::calcLocalBadness()
    change runSuitenameReport() to use "suitename -chart" (?)
    add *option* to make REMARK 42 into an unnumbered REMARK
    email new MolP tarball to Joe Moran at GSK
    **** Fix ensemble model upload history info like in 3.12****
    Possible issue with analyzing PDBs without REMARKS? (Bug report 070912)
    New F name for flips needs to reflect whether the file has any flips in it!!!
    Add flips to end of coot todo list?
    fix up fragmentfiller software page
    maintain knowledge about flips done already?
    discrepency with pdb 1gtf bond geometry total different than reported on multicrit table
    
==============================================================================
===  CHANGE LOG  =============================================================
==============================================================================
### 3.17 ###
- Switched to KiNG version 2+.
- Updates to Prekin, Reduce, suitename, Jiffiloop.
- Added direct viewing of PDB files in KiNG.
- Modified summary chart look slightly.
- Updates in cmdline scripts.
### 3.16 ###
- Interface output screen defines types of contacts.
- Added time limit to GC step so that sessions start in 1 sec or less.
- Added RNA analysis to oneline-analysis.
- Fixed bug with suite analysis not ignoring comments in suitename output.
- Changed how bond lengths and angles sort in multi-chart.
- Added Molprobity score percent rank to cmdline oneliner.
- Changed summary table resolution range output, and "Rotamer outliers" to "Poor rotamers"
- Changed bond and angle geometry goals.
- Changed html multichart to say "Poor rotamers" instead of outliers.
### 3.15 ###
- Downloading files without hydrogens trims the H off of the name.
- Adding flips to a file adds a F to the name of the file.
- Added options for turning off html or coot chart.
- Added python version of coot chart.
- Molprobity score summary box color behavior changed slightly; if percentage is better
  than 66% then the box is colored green.
- Typo in the coot script fixed.
- Fixed bugs in fragment filler code, added option for simulating a gap.
- Faster code for generating views in multikins.
- fixed bugs in reduce.
- new prekins for dealing with two char chain ids.
- option for running reduce trim added to editpdb
- Molprobity can interpret ?GET pdbCode now (for use with PDBWiki)
- Updated pdbv2.3 hashmap.
- Updated remediator.
- When clashscore is negative (pdb 2izq) no longer says clashscore is 100th percentile
- Similar patch to molprobity score

### 3.14 ###
- fixed a bug in MacOSX makefile that was preventing proper compilation of the 
  universal binaries.
- fixed MacOSX/reduce
- updated installing.html file.
- added geometry analysis to residue-analysis and oneline-analysis in cmdline/
- fixed multicritchart so geometry gets sorted by abs(sigma) instead of just the sigma (in visualize.php)
- [IWD] added session.use_only_cookies=0 to fix (?) one-shot sessions bug

### 3.13 ###
- pdb converter page now allows download without hydrogens.
- fixed bug in multi-crit table bond angle analysis.
- geometry analysis visualization from dangle has been incorporated into multicrit.
- fixed bug in pdbstat where atoms with alt confs would automatically be called pdb v3
- Suitename was updated suitename.0.3.070628
- Suitenames output on the multi-criterion chart now includes triage
  angle or 7D outlier information
- pdbcns and the remediator were updated to convert three letter
  nucleic acid names to single letter names
- The reduce het dictionaries were updated so that hydrogens aren't
  built on carboxylates
- A link was made to the Reduce download page on the reduce results
  page so that people could find the het dictionary that was being used
- REMARK 999 was changed to REMARK  40
- The spurious non-links on the lab notebook page were removed
- The perp-to-line info on the multi-crit chart now reads suspect
  instead of wrong sugar pucker, and also includes the offending torsion
- New remediator and hash table for fixing a bug where C5A on DT wasn't getting converted
  to C7.
- Tighter goal values for geometry.
- Sorts geometry outliers by sigma value
- View markers now come up as off.
- Fixed a bug where Cbeta dev on alternate confs gave wrong value on multi-chart.
- Changed MakeCootClusteredChart so it's much faster.  Results are slightly different,
  but still correct, I think.  12 minutes for ribosome as opposed to >24 hours before.
- More logical button clusters for viewing in coot.
- geometry analysis selected by default now.
- New reduce and probe do two letter chains; fixed a bug where flip page wasn't 
  recognizing the flips in the usermods and.
- Fixed a bug in loadClashlist where 2 char chain ids weren't read correctly.

### 3.12 ###
- fixed bug in warning about auto-converting to PDBv3.
- new remediator, prekin, flipkin, reduce, and probe.
- changed name of PDBconverter_bulk to remediator.
- changed warning on file upload page to be more specific about file upload modifications.
- changed model[history] so it more accurately reflects origins and modifications of file.
- added note that if no PDBv2.3 atoms are detected, assumes PDBv3 format file.
- Downgrade is (for download only).
- Updated authors on about page.
- Added references to bond and angle geometry papers on about page.
- Added "What's new" to welcome page.
- Updated site map.
- Changed Remark42 to Remark999, and updated its authors and reference.
- New Dangle to avoid doing suite analysis on DNA.
- Fixed bug in PDBv2->PDBv3 hashmap.
- Updated aacgeom_setup page so geometry and suitename are listed as beta test.

### 3.11 ###
- added fragmentfiller pages and jar (access through site map).
- have not included the libraries into svn yet, but they go in lib/fragmentfiller/.
- new versions of prekin to try to address issues with new pdb files.
- incorporated Bob's changes to analyze.php for dealing with insertion codes in RNA files.
- geometry analysis for proteins and nucleic acids added to multi-crit chart.
- [IWD] Java help page now suggests Java 1.5 (required for KiNG 2.x)
- new probe, reduce, and prekin versions for dealing with PDBv3 files.
- new pdbconverter for converting PDBv2 format to PDBv3.
- new code to check if files are old format, and convert them if necessary.
- updated references to new NAR 2007 reference.
- added warning about PDBv3 files.
- added warning to file summary page.
- added pages for converting files to PDB v2.3 format.

### 3.10 ###
- added support for icon images to linkAnyFile()
- added icons for multi-chart, -kin to analysis results page
- added suite string (+ word wrap)
- checked in the view-pdb script for Jane
- added "multichart" script for Martin Saunders at GSK
- added icon for Coot multichart
- added suitename kinemage to list of single-criterion outputs

### 3.09 ###
- fixed bug (I think): error cluster fails at 0,0,0 for certain pdbs e.g. 1lg7
- added preliminary version of new Coot to-do list with error clusters
- added architechture diagram from thesis to doc/
- updated chiropraxis.jar (newest Dangle for use w/ suitename)
- added suitename binaries (still need Universal one for Mac)
- added suitename report and multichart column to aacgeom analysis
- added count of suitename outliers to column header in multichart
- updated suitename to a universal binary for Mac
- updated Mac versions of suitename and prekin
- we now parse in the info about epsilon from Prekin
- updated chiropraxis.jar to fix "+1" parsing error in Dangle
- tweaked table sort order (native order fallback never changes direction)
- updated HTML and Coot multicharts to use new Prekin epsilon info
- fixed typo bug in findSuitenameOutliers()
- added suitename info to outlier summary table

### 3.08 ###
- added reduceTrim() function to enable downloading PDBs without H
- no-H downloads after reduce build, in file table, and aac-geom results
- added ini_set('memory_limit', -1) to oneline-analysis
- added cmdline/ssh_grid_filter to help distribute big jobs
- added -filter mode to oneline-analysis
- fixed truncation on stdin on copy for oneline-analysis
- moved ssh_grid_workers definition into a new config file
- fixed bug in file listing (came from using if instead of elseif)
- added multi-model Rama plot (PDF) for Pei et al.
- setup now runs PHP check script and opens it in a browser
- ssh_grid_filter now directs stderr to a separate file
- fixed split/joinPdbFiles() to retain model numbers when making ensembles
- added model.php :: reduceEnsemble()
- added missing "plain text" link for pdbH files
- added GUI for running Reduce on an entire ensemble
- reorganized pdbstat.php :: describePdbStats() to be more RNA-friendly
- added isReduced / isBuilt for ensembles too

### 3.07 ###
- updated mpReadfile() with provisions for very large files
- fixed bug introduced by new rotamer analysis that counted truncated sidechains
  as rotamer outliers (reported missing chi angles as NaN).
- added cmdline/residue-analysis to make CSV version of multi-chart
- changed semantics of analyze::findAllOutliers()
- many scripts in cmdline were missing mpInitEnvirons() call!
  This caused messages like "sh: prekin not found"
- added analyze::calcLocalBadness() to cluster validation flags spatially
- fixed hless.jar so it's compatible with older Javas (-target 1.3 -source 1.3)
- changed "MER" to "MolProbity Score"
- dropped "B lt 40" line from multi-chart
- fixed bug in model.php::replacePdbRemark() that dropped a line from the file.
  Thanks to Boaz Shaanan for reporting it!
- tried to make "my own MolProbity" link make more sense
- added markers for new views of clustered errors
- changed clipping depth on multikin views

### 3.06 ###
- changed file download so clicking on dir names is as good as on triangles.
  This may help keep users from getting confused about how to work it.
- added info to analysis page about how much reduce could / did improve the mdl
- fixed text formatting on reporting improvements
- added model[isUserSupplied] property, for auto-download of reduced files
- tried to make sure isUserSupplied is propagated when new models are made
- added special "done" page for Reduce that prompts for file download
- added unlink() commands for temp files in code that reports improvements
- changed "angles" to "chi angles" in rotamer column of multichart for JSR
- added rotamer naming to runRotamer() -- required a new chiropraxis.jar
- added rotamer names to multi-crit chart output (formatting could be better)

### 3.05 ###
- fixed up some rare weirdnesses about guessing the user's IP address
- multi-chart generation is not too slow:  only 17 sec for the ribosome
- added checkbox for "outliers only" to be listed in multi-chart
- boosted Java memory limits to avoid OOMEx on 1S72
- changed Rama sort values so now *all* outliers sort to the top
- worked out UI logic for multikin views vs. running multichart stuff
- wrote linkAnyFile() to replace and augment linkKinemage()
- changed notebook entry display in upload_pdb_done to match generic_done
- reformatted links in main analysis results page
- fixed A / A-ring mixup in multichart
- made pdbstat.php more able to recognize CNS-style REMARKs
- updated KiNG in hopes of not leaking applet memory
- updated Probe, Reduce, Cluster to universal binaries by WBA
- added mpSerialize() and mpUnserialize() to solve float bugs in multi-table:
    CBdev sort bug, PHP only, 1E8C
    Rama sort bug, PHP only, 1AS4 (B 390 ASN)
    High B calculations sometimes produces "INF" (1ECF A 143 ARG; 1as4 a 274 arg)
- added "feature" CSS class for divs -- to be used in tutorial pages
- tweaked around with the colors a lot and ended up back at the same place
- started working on helper_xray page
- upgraded to a newer Universal binary of Probe 2.11.061018
- commented out NMR multi-chart, for now
- added helper pages for NMR, rebuilding, and kinemages
- added page for downloading MolProbity
- added LICENSE file (BSD license)
- updated KiNG to newest parallel coords code for use by Bodan
- added debugging code to job_progress for cases where PHP syntax error causes
  the background job to abort before it ever begins running.

### 3.04 ###
- added public_html/data/index.html to thwart directory listing attacks
- added HTML "style guide" page as documentation for our CSS stylesheet
- fixed presumed PDBCNS typo bug in handling nucleic acids (\\047 vs. \047)
- failed event_page stuff no longer wipes out session data
- fixed calls to curl_errno() in model::getEdsMap()
- fixed $req/$_REQUEST assignment bug in welcome.php::onUploadOrFetch()
- added CSS style guide HTML page
- fixed race condition in launchBackground() that could truncate the session
  data file before the background process could read it in, leading to
  random failure of the background process.
- added per-residue views in the multikin
- minor formatting changes to job_progress, generic_done
- changed "Analyzed ___" to "Analysis output: ___"
- added -code to Prekin in Flipkin to remove name of tmp file from pointIDs
- fixed map upload code to allow gzipped maps
- Reduce update fixed one longstanding bug (wrong protonation for 1gca His 152)
- changed from using CCP4 maps for fetch to using O maps (.omap.gz) [JSR]
- tried getting rid of subfolders -- file become to messy
- added JavaScript to the file browser; use JS if enabled, page refresh if not
- changed "cb2d" kinemage name to more intuitive "cbetadev"
- moved fix.flips file from coordinates dir to raw_data
- multichart can now be sorted by residue type as well as residue number
- rewrote usage_history to use much less memory (I hope)

### 3.03 ###
- many upgrades to page handling from PhD Posters work: setContext(),
  makeEventURL/Form(), makeDelegateObject(), mpControllerDie()
- removed $req from all event handler functions
- removed extra arguments from makeEventForm() -- file upload, onsubmit=
- removed unused $arg parameters
- moved onReturn() handler into BasicDelegate
- moved mpPageHeader/Footer() into BasicDelegate
- changed onNavBarCall/Goto to onCall/Goto
- oops! mpPageHeader/Footer() must (also) be in core.php; added $headContent tho
- ... and after all that, MolProbity still seems to work (mostly)
- updated page template in doc/
- added Google Analytics tracking code
- removed link to old MolProbity, which is no longer functioning
- updated developer documentation on MolProbity event system
- added cmdline/multichart-coot for Mitch Miller
- moved Google Analytics code to be part of the config files
- fixed bug seen for 1CEQ where insertion codes hit res types (BGLN, EASN)
- closed a major security vulnerability that allowed upload of PHP files!
- updated documentation to mention this type of hole

### 3.02 ###
- updated effective resolution code, including a new percentile rank score
- add H no longer suggests making flipkins if no protein is present
- added tool to convert apparent NMR ensemble to a biological unit file
  (i.e. convert MODELs to chains), because there's no way to recognize an
  uploaded biological unit file.
- updated effective resolution code and library to use floors for rotamers
  and Ramachandran -- no reward past "expected" values.
- updated Mac Reduce to 3.02, the Snoeyink lab version
- updated Mac and Linux Reduce to 3.03
- removed -limit flag temporarily, until it gets re-implemented in Reduce
- AL-Fs patch for Reduce in His flipkins (Mac + Linux)

### 3.01 ###
- we've been in production mode for a long time now, so I changed to a more
  normal version numbering scheme.
- created Scheme script for COOT that lists all multi-crit outliers.
- updated numbers used in effective resolution calculation.
- made pdbstat find other nuc. acid H; changed criteria for has_most_H

### 3beta29 ###
- added Life, the Universe, and Everything question to april_fools.php
- updated Linux prekin to stop seg fault on long filenames
- increased -Max# for cluster within clashlist, to fix bug running 1iw7
- started work on JavaScript sortable table; it doesn't work so far
- finally got JS sortable table working; it's only used for the static copy
- fixed sortable table for IE -- hasAttribute() wasn't supported by IE
- 1BKR has abnormal "INF" values for waters 1085, 1087
  This is a PHP serialize bug in 4.4.1 (and earlier?); fixed in/before 5.1.2
  No known work-arounds that I can find.
- clashlist, line 99: changed "single" to "-single"; makes no difference?
- added clashlist.annotated.sh with comments on how it works
- added process ID ($$) to clashlist tmpfiles to avoid collisions
- copied annotated clashlist over the original
- updated chiropraxis.jar (Rama PDF pct fav,out and citation)

### 3beta28 ###
- added temporary work-around to out-of-memory during usage history
- added more stats about memory usage
- compiled (non-static) Intel binaries for Mac: probe, prekin, gawk
- combined them with old versions to get universal binaries (using lipo -create)
- "effective resolution" calc is now running -- perhaps even correctly
- added PhD Posters link in page footer; check this w/ Dave and Jane
- added "banner" for msgs about system going down, etc.
- added effective resolution to oneline-analysis script.
  Top500 shows little correlation to actual resolution -- is there a bug?
- multikin: {Virtual BB} master needed to be turned on for nuc acid structures
- extended file lifetime on main server to 3 days (for bug report use)
- fixed local setup bug (missing symlink for awk used system awk)
- upped Java memory for rotamer eval. b/c 1htq caused OOMErr
- removed PhD Posters link before checking w/ D and J; we have enough business
- added electron density, restriction sites questions for April Fool's

### 3beta27 ###
- updated Probe to 060129
- added "dots" master for (original) Probe dots in the multikin
- changed small/wide contact masters in multikin to "vdw contact"
- unchecked JNB's chart by default and moved the "alpha test" notice around it
- added "Models colored by" to B-factor/occupancy boxes in analysis setup
- changed logic of Interface selections because (prot|nuca|wat|het) !== (all)
  That is, some atoms were excluded even when all boxes were checked.
  Now we EXclude classes that are unchecked rather than INcluding checked ones.
- updated Prekin (new P perp to line code)
- updated Prekin to 6.43.060216; fixes -range bug from last update
- reworked interface selections once more -- does this now cover all cases?
- added link back to the old MolProbity site
- added note to welcome.php about not using the back button
- no back button warning now has a "no harm done" message too
- fixed upload error bug (no $this)
- added "don't see anything?" link to thumbnail on upload page
- bug fix: notebook edit deletes icon
- added join-models cmdline tool
- added distance checks to Java phi/psi code -- caused spurious Rama outliers
  at internal chain breaks.

### 3beta26 ###
- updated KiNG to 1.49
- changed table highlight color to coordinate better with site color scheme
- tried to make pdbstat do a better job of getting R and Rfree; seems to help
- added -noXXX flags to oneline-analysis
- KiNG viewer page tries to auto-widen the browser window using JavaScript
- Craig Jolley complains that multicrit chart generation is slow for large
  structures. He's right: his 7113 residues take 57 sec on my laptop. However,
  the serialized PHP array is 5 Mb (!) and doing much better would (I think)
  take a LOT of work. Sorry, Craig -- not worth it.
- removed "primaryDownloads" feature entirely. Just let user pick, instead.
- added ability to set/change resolution while editing a PDB file
- tried yet another layout for the upload boxes

### 3beta25 ###
- upgraded prekin again (051130/051129)
- changed output of bad rotamers so they're all under one subgroup
- updated prekin again (051201) - fixes weird masters bug, adds -version
- put probe/reduce/prekin version numbers in multikin text (DCR/JSR)
- experimental JavaScript popup menu interface for download from welcome.php
- that's not great -- still too messy
- tried plugging in the folding file browser; with a few tweaks, this isn't bad
- added interface for deleting chains from a PDB file
- fixed welcome.php to show list of results even if there aren't any models
- added $_SERVER[HTTP_REFERER] to log
- extended cmdline/oneline-analysis to include clashscore percentile rankings
- collapsed runAnalysis() and runAnalysis2() back into one function
- added mpTempfile() and converted all of lib/ and cmdline/ to use it
- added logging of user pathway thru pages on the system
- added scripts that feed Java 3D layout code for visualizing user paths
- removed debugging output from model.php::remapSegIDs that junked up output
- made oneline-analysis ignore segIDs (junk segIDs on Top500)
- added numRama to oneline-analysis (could get this by adding the 3 categories)
- added script for plotting server load (using uptime)

### 3beta24 ###
- changed "reduce_done" to "generic_done" in reduce_setup.php: this was
  breaking the "Add missing H only" command.
- multikin checkboxes for B-factor and occupancy were separated for JSR
- various cosmetic fixes for JSR
- fixed JS model selection form on welcome.php to work in Firefox
- modified session expiration date tracking in lib/sessions.php
  Added MP_SESSION_LIFETIME_EXT to config; use it in pages/save_session
  Fixed admin/show_sessions to use the new last-touched vs. TTL data
- changed setup.sh to avoid clobbering config.php
- left defaults in config.php.sample but commented them out to start
- moved browser and java warnings to the bottom of welcome.php;
  omitted the Java detect for Mozilla-like browsers on OS X (false negative)
- stopped adding "ens_" or "_clean" to uploads, except for uploads that really
  did need something cleaned up in them.
- can now get maps from welcome without (re-)getting model (JSR)
- upgraded to Prekin 6.42.051128
- reworked code for multi-criterion kinemages using new Prekin features
- modified pbdstat to count waters
- fixed logging cmds in aacgeom_setup that broke when form fields were renamed
- updated KiNG (automatically removes unused groups/masters)
- added repeating headers to usage_history.php (needed for long histories!)

### 3beta23 ###
- fixed LONG-STANDING SECURITY EXPLOIT that has affected ALL previous versions
  of MolProbity. Because molprobity.log was not protected from access via HTTP,
  someone who knew it was there could pull it up in their browser and thereby
  get a list of EVERYONE's session IDs. This would allow access to basically
  all user files on the system. This has been fixed by moving the log to
  the feedback/ folder instead.
- rearranged list of browser usage into a table format instead
- added Gary's list of interacting atoms to interface vis pages
- from welcome.php, skip over model-choice segment of interface vis
- fixed bug in bot filtering for usage_history.php, listed more bot types
- totally rearranged the main validation interface, which required making
  runAnalysis2() and makeMulticritKin2(). Old versions should go away later.

### 3beta22 ###
- merged upload_pdb_setup.php and upload_other_setup.php. This simplifies the
  site overall, and the overwhelmingness of the UI is less important b/c 90% of
  uploads will now happen from welcome.php
- changed "Advanced options" styles to be less obtrusive
- added upload options (e.g. EDS fetch) to welcome.php using JavaScript
- fixed title on makekin_done.php; was "AAC and geometry"
- reformatted notebook entries on welcome.php to be reverse chronological
- added simple popup viewer for lab notebook entries
- greatly simplified displayModel/EnsembleTools() code; added small icons
- cleared out unused files from public_html/img
- matched lab notebook TOC to the TOC in welcome.php
- added (optional) icons to notebook entries to facilitate user recognition
- changed feedback system to be able to use SMTP; it seems to work
- made splitModelsNMR() an alias for splitPdbModels(); marked it as deprecated.
- made KiNG window sizes a bit larger all around. Unfortunately, percentage
  sizing doesn't work reliably in either Safari or Firefox. Too bad.
- changed lab notebook titles to "Verbed name-of-tool for PDB file" style
- changed "AAC analysis/analyses" to "Analyze AAC"
- simplified pages/ with a generic "done" page that uses notebook entry title
  All remaining specialty "done" pages have additional special code in them.

### 3beta21 ###
- changed config.php to config.php.defaults; modified setup.sh appropriately
- made config.php ignored by svn
- created defaults.php; renamed config.php.defaults to .sample; fixed setup.sh
- elaborated on Firefox/Java setup instructions
- fixed missing master for Rama outliers in multi model multi kin
- added cmdline/remark42 for Wolfram Tempel
- added ALPHA TEST notice to JNB's NMR tools page
- superficial rename of welcome.php to Main Page (JSR)
  Didn't bother to rename the file(s) b/c she may change her mind later!
- made welcome.php easier to navigate by breaking functions into labeled boxes
  and regrouping functionalities.
- file upload box can now handle any type of upload via a dropdown menu
- simplified and re-ordered navigation sidebar

### 3beta20 ###
- can download the biol unit from the PDB, but it comes as MODELs
  can't just strip MODEL/ENDMDL -- breaks Rama, rota, other Java, multi chart...
- added MODEL-to-chain mapping for download of biological units
- fixed bogus "0 bad rotamers" in multi-chart header (a typo)
- removed hits by bots and crawlers from the usage stats pages
- wrote mpReadfile() to deal with readfile() bug in PHP 5.0.x for 2Mb+ files
- created shell for doing multi-model multi kins
- got a "rough draft" of the multi-model multi kin working
- fixed up some cmd line scripts: analyze-nobuild, oneline-analysis
- removed lib::addModel()
- added cmdline/multikin for Vince
- made summary stats table more like remark 42
- headers now repeat in multichart every 20 rows
- SSWING is now accessible only from the sitemap page and is marked "alpha"
- listed multi-model multikin on sitemap page
- improved formatting of chi angles in multichart (n/a is now left off, not 0)
- NOT REPR: JSR: Treat MSE as MET in Rota, Rama analysis (1NSF)
- changed loadRotamer() to be keyed on CNIT instead of integer index
- changed loadRamachandran() to be keyed on CNIT instead of integer index
- updated Java Ramachandran code in hless.jar to handle MSE
- fixed a bug in loadRotamers() introduced by the recent chi angle "fix"

### 3beta19 ###
- added directory check to setup.sh
- added getEdsMap() to lib/model.php
- added UI for fetching maps from EDS
- added backend for fetching maps, including progress steps and notebook entry
- cosmetic changes to welcome.php
- config now checks for fopen wrappers enabled
- updated sitemap.php (all listed features are now implemented!)
- added $_SESSION[ensembles] for managing NMR files
- in model.php, wrote splitPdbModels() and joinPdbModels()
- added IE specific headers for our tricky ZIP download in welcome and browser
- PHP redirected downloads have been tested against IE 6.0 / WinXP
- changed welcome.php to use lastUsedModelID safely (as model OR ensemble)
- debugged addModelOrEnsemble() and converted jobs/addmodel.php to use it
- added check all/clear all javascript to the file downloads page
- wrote the installation instructions
- standard page header is now INDEX,NOFOLLOW (so Google will see the front pg)
- tried a warning page for users who exceed disk space; it's too expensive
- improved Start Over link on logout and added link to Kinemage Home Page
- simplified getPdb/NdbModel() by using copy() with URLs

### 3beta18 ###
- changing the REFRESH tags fixes MP3 to work with IE6 ... mostly ;)
  See "Other people's bugs", above.
- started logging the user-agent (browser identification) strings on sess start
- added browser detection code from the web
- improved usage_history.php, added browser usage statistics
- added human-readable error messages to upload error pages;
  modified welcome.php's upload form to work even when cmd is not set.
- added JavaScript warning about using back button/multiple windows (bad events)
- changed JS to alert div at start of page
- added browser (MSIE) and no-Java (via JS) warnings to welcome.php
- updated Java instructions page from MP2 and linked it from welcome + viewking
- session view page now sorts by time-to-live by default
- fixed typo bug in browser identification code
- chose not to create "robots.txt" b/c this must go in webserver root dir
- added tests for bots/crawlers to browsers.php
- added extra realpath() calls in viewtext/table to make sure check succeeds
- added makeZipForFiles() to core.php
- added download of ZIP of "popular" files from welcome.php
- added configurable ZIP downloads for the file browser page

### 3beta17 ###
- started on new welcome "all in wonder" page
- changed $delegate declarations before classes to be done dynamically from
  within index.php as $delg_class = basename(...); $delegate = new $delg_class;
  This required renaming all the classes in pages/ and updating devel docs.
- continued development of new welcome page including working model selector
  with JavaScript autosubmit and list of tools with pretty new icons.
- moved global var to class var for file_browser.php
- improved formatting on welcome page; still needs work
- added file downloads and lab notebook entries to welcome page for curr model
- added "primaryDownloads" list per model
- added "has_most_H" field to PDB-stat results
- merged welcome pages back together into one
- cleaned up site map to be more bare-bones again
- misc improvements to welcome page code
- added debugging page that displays $_SESSION, linked from admin pages
- added icons for making kinemages and interface contacts
- verified that IE 6.x doesn't work with MP3 at all, presumably due to weird
  caching behavior -- e.g. uploads just leave the user at the Welcome page.
- added anti-caching header() calls to mpInitEnvirons() to combat IE6 bug.
- not a page-caching issue. It looks like the background job never gets started,
  because the system/errors file is never even created.
- actually, the session ID doesn't get preserved during background job
  processing -- IE doesn't like index.php?SESSION_ID... in the REFRESH tag.
- added URL=... to all REFRESH tags. Google says this should fix the IE bug.
- commented anti-caching commands out again. If it ain't broke...

### 3beta16 ###
- fixed bug in logout_destroy.php that kept the session ID out of the log file
- added sorting and "no sessions" msg to show_sessions.php
- changed config.php and core.php to auto-detect server timezone (usually?)
- added page for monitoring usage history based on molprobity.log
- added links between the various admin/ scripts
- housecleaning in check_config.php for more sensible order
- added much more detail for 'aacgeom' log messages
- upgraded Prekin to do -pperptobase to -pperptoline change for Laura
- made link to "alpha test" more visible on kinemage
- usage history now has selectable columns for detailed breakdown
- added "clashes-with" information to clashlist parser and multichart
- moved most HTML formatting for multichart in lib/sortable_table.php
- created multichart snapshot in charts/, main data in raw_data/ -- is this OK?
- rebuilt glass with Michael as the new alpha-test host of MolProbity3

### 3beta15 ###
- clash stats for score B lt 40 are now done correctly rather than against the
  raw clashscore, which was producing an unfair negative bias for all structures
- made explanation of percentile rank more upbeat: "100 is best...; 0 is worst"
- experimental: gzip large multikins (see core.php, analyze.php in lib/)
- file browser: links for .kin.gz, .gz, .tgz, .zip, .html
- core.php now turns off E_NOTICE, which should save reconfiguring newer PHPs.
- cleaned up multikin gzip code -- it seems to work, so we'll keep it for now.
- removed the (ugly) gray clock face from behind the spinning SOD "progress bar"
- moved makeSummaryStatsTable() from analyze.php to visualize.php
- moved to PHP 5.0.4 due to a printf() bug prior to 4.3.7
- wrote makeRemark42(), which was affected by said printf() bug
- wrote replacePdbRemark() to insert/update REMARK 42 records
- added interface for REMARK 42 in AAC/geom. pages
- fixed hless bug that threw NPEx when drawing 2SOD Rama outliers
- linked flipkins (if avail.) to Reduce notebook page
- removed "highlighted" kinemage view from downloads page
- added some extra version checking to check_config.php
- added res hiB to multichart for JSR
- made multicriterion chart sortable by storing as an encoded PHP array and
  writing public_html/viewtable.php to display it and sort it.
- updated analyze.php::runAnalysis() and aacgeom_setup.php for new multichart.
- added support for .table files in file_browser.php

### 3beta14 ###
- added "start" parameter to allow starting other places besides welcome page
- following bug has been fixed by update to Prekin 6.40.050310:
    Prekin: excess rota/rama outliers from too-long -range statement (?)
    test cases: 1FFY, 1XFH, 3AL1 (used to; doesn't any more)
- PdbStat now does chain identity as substring test rather than identity
- changed "AAC" to "All-atom contacts" (JSR); required changing minornav SPANs
  to DIVs in the navigation side bar. Safari still renders them correctly...
- changed viewer close button: When done [close this window]
- Multikin text: "Statistics for first FILE only..." becomes MODEL
- multikin: "mc/sc alts" to "mc/sc alt confs" and come up turned off
- NMR model splitter now puts model number first, not last (JSR)
- AAC/geom: "Output Formats" section now mentions extra plots/kins/lists (JSR)
- restored makeBfactorRibbons() function, though it's not in the multikin yet.
- fixed: Sort by Outliers moves hets to top (het chain = _) (1U4B)
- added gzip to list of required programs anticipating compression of lg. files
- created overview flowchart for interaction of different UIs with lib/*
- seriously reworked the multi-criterion kinemage for single-model mode
- changed colors, esp. Rama markers to thick, green, half-Ca bonds.
- changed rainbow ribbons back to B-factor ribbons for single models
- phi,psi output in multi-crit list now truncated to one decimal place
- put (some) summary stats at top of multichart
- put stats summary table at top, too
- factored summary stats table out of analyze.php:runAnalysis()
- fixed CBdev AWK script so master works for all subunits (test case: 2SOD)

### 3beta13 ###
- added check in index.php that a delegate page is actually specified
- simplified AAC/geom interface based on feedback from JSR
- made generation of B-ribbons, alt confs optional in multikin
- replaced B-ribbons with B- and Q-color scale renderings plus rainbow ribbons
- made summary stats optional
- color-coded summary stats results as good/fair/poor (traffic light)
- swapped positions of AAC/geom and SSWING icons on sitemap at Jane's request
- removed stand-alone 3-D Cbeta deviations kinemage b/c it appears in the multi
- implemented base-P outliers in multi kin, summary stats, multi-chart
- enabled sorting of multi-chart by baseP outliers
- updated Probe and Prekin (v050121) binaries for OS X, Linux
- added Welcome page (mostly without working links); made index.php start there.
- reorganized formatting of side links in lib/core.php
- added extra citation (NAR) to help/about
- report # H added/deleted/adjusted when running Reduce (build or nobuild)
- excess rota/rama multicrit outliers traced to Prekin (1FFY, 1XFH, 3AL1)
- changed preferred fonts to Arial and Optima
- created draft version of x-ray helper page
- can't catch back-button event to warn user not to...
- added color code for occupancy greater than 1.00
- turn on masters for alpha/beta/coil

### 3beta12 ###
- created command line scripts: reduce-build, reduce-nobuild, split-models
- explicit resolution range used for clashscore pct on analysis results page
- use "-nobuild" in descriptions even though it doesn't exist (yet)
- converted oneline-batch.php into oneline-analysis script for modifed sess structure
- changed model name finder to use underscore if model name ends in number
- sitemap now lists available models
- imported complex Probe runs (interface visualization) from MolProbity2
- actually moved MolProbity repository to new cluster
- enabled Probe commands for interface contacts
- split "Protein,DNA,RNA" selection into protein vs. nucleic acids
- added -DROP option for interface contacts
- fixed bug in describePdbStats() that never saw hydrogens
- modified cs-rank.awk to compare no-resol structures to entire c-score database
- made some navigation icons for sitemap; font is Helvetica Neue, 12pt, 75 Bold
- allowed makeEventForm() to accept a onSubmit cmd
- AAC page now warns about missing H only during form submit
- simplified JavaScripts on AAC page
- slightly changed wording, code for other no-H warnings
- replaced PHP_SELF with basename(PHP_SELF) b/c of kinemage-arachne URL rewrite
- added extra info to admin/check_config.php

### 3beta11 ###
- moved timezone setting to top of lab notebook page
- added noe-display to bin/
- removed public_html/job_progress.php, because it's been superceded.
- wrote document about the event/delegate model for the GUI
- added new "wait" graphic based on 2SOD (clock face)
- added count of mc/sc alternates in pdbstat
- added note about percentile ranking for clashscores in multi-output
- changed sc alts to cyan to complement the blue normal sc
- added @onewidth to multikins; turned stuff off/on with masters
- updated KiNG to support new on/off (point)masters
- finished doc on adding pages to MolProbity
- simple kins page works, but needs more detail (N--C color, etc)
- implemented HTML snapshots of lab notebook for later reference
- system now tracks "last used" model to give more continuity
- changed to high-penalty -build for "nobuild" Reduce runs
- simple kins page now makes sure there are actually models to act on
- AAC setup now syncs with pre-selected file on page load
- simple kins supports rainbow coloring and CPK atom markers
- reworked animated 2SOD "clock" graphic for job waits
- added more description to Site Map and linked to Feedback and bugs
- fixed subgroups in multicriterion kinemage
- added @onewidth to flipkin
- updated help pages to match site style sheet
- wrote "reduce-build" and "reduce-nobuild" command line scripts

### 3beta10 ###
- continued work on aac/geom analysis, rendering results as a notebook entry
- removed options for half-bond colors and atom-markers; this should be part
  of the simple kins. User can append later using URL filesystem!
- rewrote multi-crit kin function borrowing ideas from Jeremy's NMR kins
- not a bug: $_SESSION[hetdict] is used to set an environment var instead of directly
- added lib/models.php::splitModelNMR() based on JNB's code
- enabled upload and splitting of multiple NMR models
- fixed pdbstat() to not find "RESOLUTION. NOT APPLICABLE.  SEE REMARK 4."
  as a resolution of 4.0 A
- modified model splitter to REMARK out MODEL and ENDMDL lines
- improved reduce_choose page for case of no flips
- removed unneeded old pages from public_html (~r88)
- we now display comparison to run-of-PDB (SCOP 2000) for clash score
- added sortable multicriterion chart to aac/geom analysis page
  Sorting on the fly is complicated b/c of formatting issues. We could consider
  a page that lets you regenerate that chart to simulate dynamic sorting.
- fixed sorting bug for unset Rama, rota values appearing as zero
- did a little more work on check_config.php
- eliminated symlink-to-binary setup for pdbcns
- eliminated symlink-to-binary setup for other executables, except gawk on OS X
- removed unneeded statements from makelinks.sh
- renamed makelinks.sh to setup.sh
- updated reduce/prekin/probe on linux, os x
- updated linux SSWING binaries

### 3beta9 ###
- changed PDB upload back to stacked, rather than 2 column layout -- it's more
  consistent and visually cleaner. Fetch by ID is de-emphasized a bit, OK.
- added JavaMage thumbnail to notebook entry for PDB file upload/fetch.
- updated binaries for SSwing
- started building interface to SSWING again
- allow user to choose files with no H for SSWING, but warn them first w JScript
- finished SSwing interface, minus info about which sc really changed conf
- improved formatting/layout on Reduce's "choose" page
- added CSS rules for help text divs
- help text for setup pages is now grayed out; could change background instead.
- simplified sitemap.php by using onNavBarCall() instead of custom handlers.
- started AAC/geometry analysis UI; added JavaScript smarts to it
- finished JavaScript work in aacgeom_setup.php; got radios disabled
- JavaScript tested against several browsers, and all are fine:
  Mac: Safari 1.2.3; Mozilla 1.6; IE 5.2
  Linux: Mozilla 1.0.1, 1.2.1; Konqueror 2.2.2 (disable broken), 3.1-15
- added debug switch to sswing (-d); added debug output to sswing.php
- "Fast search" checkbox for sswing is now actually connected to something
- fixed setProgress() / getProgressTasks() bug (variable not marked global)
- added progress marker for making -cass kinemage during upload

### 3beta8 ###
- started work on the Reduce pathway; got the setup page done.
- swapped "Close" btn to left side of viewtext.php
- got reduce-build and reduce-nobuild jobs running, but no results page yet
- wrote save_kinemage.php that works with KiNG to allow user to save kins online
- changed reduce_setup so it's still possible to re-Reduce files with Hs
- added "choose" and "done" pages to Reduce cycle; still needs polishing
- I think the Reduce cycle is finally finished...
- started on feedback system
- removed -noadjust flag from Reduce -nobuild cmd lines; apparently causes
  problems with His protonation...
- refactored model-creation code into createModel()
- file upload now talks about PDB files, not "models"
- notebook can take bar-separated list of modelIDs instead of a single ID
- reduce-nobuild now generates a new model entry, and talks about file names
- fixed getVisitorIP() b/c $HTTP_* aren't registered by default anymore
- finished feedback pages
- broke upload pages into two parts
- added kinemage upload and viewing

### 3beta7 ###
- added forms for ED map and het dictionary upload
- moved session, errors, lifetime, labbook, progress to system/ folder.
- fixed admin/show_sessions.php to work with the new MVC/event model.
  (It didn't require changes for "session" moving. Yay for encapsulation!)
- removed dependency on event_page.php from sessions.php
  The MVC/event model should be a purely optional layer on top of the session
  functions; the session functions should be usable with the rest of it.
- made a really nice file view/download page with collapsible folders
- D&J decided on a file-type-based folder system for organizing user files
- added logout handler and save-session handler
- updated the site map with download, logout, save session pages
- changed to MP_DIR_xxxx system for naming subdirectories, b/c Dave and Jane
  aren't sure what they want yet, and it will probably change later.
- file upload functions now group files into directories by file type
- KiNG viewing now looks in the (new) right place for electron density
- many things that used the old model data structure will be broken, b/c
  we've removed the per-model directories (and the corresponding URLs).

### 3beta6 ###
- created MVC/delegate architecture in public_thml/index.php
  and lib/event-page.php. See those files for documentation.
- added clearstatcache() to session read, just in case (ala lab notebook)
- created "site map" page based on the MVC/events model
- ported notebook view/edit pages to MVC/events model
- remodeled core.php:mpNavigationBar() to work with MVC/events
- added base class for all MVC/events delegate classes (works w/ nav bar)
*******************************************************************************
* NON-MVC MOLPROBITY3 IS NOW OFFICIALLY BROKEN. FURTHER WORK MUST USE MVC!    *
*******************************************************************************
- BUGFIX: mpSessReadOnly() did nothing b/c variable was not marked "global"
- background event launching has been converted to MVC/event system.
  This is untested thus far but appears to be working.
- removed lib/labbook.php:openWithEdit() because old pages are broken anyway.
- considered allowing return values and handlers for pageCall() / pageReturn()
  Ended up deciding against it -- see index.php for the complete argument.
- modified style sheet to get a more subtle look
- finished file upload (except for error handling)
- fixed 'abort' function in new job progress page
- got the fetch from PDB / NDB function working
- added upload error handling
- created initial pass on the site map's formatting
  
### 3beta5 ###
- traced mysterious labbook load/save bug from beta3 to missing clearstatcache()
- used that fixed bug to simplify improve_sswing_done.php; seems to still work
- running 1S72 (lg ribosomal subunit) requires ~40 min (?) and 450 Mb on arachne
- running 1S5L (photosystem II) on arachne takes 169 min and 62% of 200 Mb
  (that's without an AAC kin). The 102 Mb multicrit kin won't load in web KiNG.

### 3beta4 ###
- changed sswing runs to use idealPDB.pdb as the output, rather than stdout
  stdout formatting was not consistent enough to rely on
* Shuren tells me this is only a starting point for optimization. There is not
* yet any file that just contains the final sidechain conformation!
- no longer allow submission of SSWING form if no maps are present
- new models from PDB/NDB are now given lowercase names
  this plays better with the xxxH.pdb convention for Reduce naming
- added "preview" links to analysis run
- added session-overwrite safeguards to viewking.php for previewing use
- created infrastructure for setting a per-user het dictionary in $_SESSION
- created upload pages for custom het dictionary
- got modified SSWING from Shuren, now read result from sidechainPDB.pdb
- changed lib/analyze to make some quick kins first to keep user entertained

### 3beta3 ###
- reworked lab notebook functions to allow automagicical entries and editing
- added automatic notebook entry for file upload
- made Analyze to Reduce -build link go to the setup page instead of improve_tab
- improved the job_progress Abort function
- added framework code to support lab book in analyze_display
  some odd hacks were required to get the right behavior -- I'd love
  to find a more elegant way of doing something like this.
- notebook edit page now protects HTML special chars properly upon save
- analysis page now generates reasonable notebook entries
- added NAR reference to home_tab.php
- added automatic labbook entry for Reduce -build
- added automatic entry for SSWING, with more ugly hacks.

### 3beta2 ###
- added timing notes to SSWING setup page
- links now highlight on hover using CSS A:hover
- changed feedback code to use \r\n instead of \n for new lines.
  This eliminates the error message, but mail still doesn't get delivered?
- now caching user feedback on disk in addition to trying to send mail.
  Not elegant, but ensures nothing is lost.
  Feedback page now doesn't report failure except in the system log.
- changed CSS to remove border from link images (needed in Mozilla)
- fixed potential div-by-0 bug in analyze_display (lines 51,59,etc)
- added display of B lt 40 clashscore
- made some timing improvements in job_progress: slows down for long jobs
- feedback page now asks user to not delete files
- fixed bugs on notebook_edit and added note about using HTML tags
- added "Close" button to kin viewing page
- added controls to set size of KiNG window
- changed "Get input models" to the more accurate "Get input files"
- added CSS for syntax highlighting of kinemage files; updated king.jar
- added syntax highlighted view of kinemage files
- fixed a file-download annoyance: now files and dirs sort separately
- added file viewing links to download page; combined viewhtml and viewtext
- fixed addmodel.php (upload system) to rename files with "weird" names
- patched filesystem-access security hole in viewtext.php
- patched filesystem-overwrite security hole in ED Map Upload system
- fixed addModel() to protect from case-insensitive file systems like OS X
- removed segID remapping from reduceNoBuild() b/c it's not appropriate
- segID remapping is now done immediately on upload
- user is now warned if segID remapping occurred
- this avoids the bug from Top500/1aacH.pdb, which had its H in segment 'new '

### 3beta1 ###
- first beta release for testing inside the Richardson lab
- rotamers weren't showing up in multi-crit kin due to a stray "s" before -scope
  Thus the error msg "Prekin can't open s-scope". Probably due to jEdit save bug

### 3pre17 ###
- decided all logging calls should be from public_html/* pages b/c we want to
  know about web traffic, not about scripted use by local users.
- inserted logging calls on many public_html pages
- got basic SSWING function working! Yay!!
- added "X minutes and Y seconds" formatting to job update page
- made "download all files" always avail. from files page
- added ED map support to viewking.php
- added custom views to the SSWING kinemage
- SSWING now creates a new model for the changes to go into
- added basic dots to SSWING kins
- added link to EDS (still need auto map download)
- modified sswing (Perl) to delete all files it creates; we use stdout only.
- changed sidechain colors for SSWING kins
- made sure proper SSWING binaries are in place for both OS X and Linux
- user can now accept or reject changes made by SSWING

### 3pre16 ###
- cleaned up lab notebook formatting a bit. Still needs better styling.
- wrote text for log out and save session pages
- wrote text for Reduce setup page
- formatted Asn/Gln/His checkbox table with MP_XXX colors
- wrote some text for after you've run Reduce
- started working on Feedback system
- completed code for Feedback system, seems to work but needs live testing
- added B-vals in pt ID and blackbody colored B-factor ribbons to multi kin
- added explanatory text to analysis setup page
- hacked together a crude multi-criterion chart
- added electron density map upload
- improved the analysis results page to make it more usable.
- added PDBSTAT text to the multicrit text window
- added makeZipForSession() to core.php
- file download page can now package things into ZIPs
- started building in mechanisms for running SSWING
- Subversion repository got screwed again!!
- upgraded to svn 1.1.0-rc1 and rebuilt repos with "fsfs" filesystem
  (fresh checkout is not needed if new repos is loaded from dump of old one)
- added SSWING binaries and scripts

### 3pre15 ###
- elaborated oneline-batch to handle running Reduce -nobuild too
  Still needs some work to really be clean.
- added listDir() to core.php
- used listDir() to clean up the weird problems with oneline-batch.php -reduce
- got new Prekin, Reduce
- learned the hard way to not do Subversion work inside sym link directories.
  I could recover by doing dump/load and a clean checkout, but scary...
- got new Reduce, Prekin binaries for the Mac
- makelinks.sh now also creates molprobity.log and sets its permissions
- still seems to work on OS X (whew...)
- changed limit for Reduce to 100,000
- added notification of disk space used
- fixed coloring issue for download table
- added robot-exclustion META to header
- cleaned up home_tab for public viewing
- added ALERT divs to home_tab and improve_reduce_choose
- commented out non-functioning placeholders in upload_tab
- dropped Reduce -limit to 10,000 b/c now 1SBP flipkins take a LONG time
- background jobs now record their PID and user can abort them (POSIX only)
  
### 3pre14 ###
- refactored multicriterion kin creation into makeMulticritKin()
- created outlier-res-batch.php to do analysis for JNB
- we now use dirname(__FILE__) instead of getcwd() to define MP3 install dir
  This means e.g. that cmdline/ scripts can be run from any directory
- removed useless Cb dev statistics
- modified pdbstat regex's again for R/Rfree to catch data from TNT
- changed resolution regex to pick up on resolutions like "2." and ".54"
- changed fixH.pdb suffix to Hfix.pdb because its easier to get right
- modified clashlist to avoid the divide-by-zero error for no atoms B < 40
- made the default Linux bin dir "linux/", symlinked it to linux-rh73/
- added Perl to the list of external programs (for Mike's scripts)
- removed defunct Cb stats from oneline-batch.php
- reworked decodeReduceUsermods() to deal with rare cases that we were failing on
- added a countflips.php command-line script to checking on what Reduce did

### 3pre13 ###
- created a tmp/ directory just for use by MolProbity (file uploads, ZIPs, etc)
- allowed user to pull files from PDB
- reworked pdbstat() to be slightly more efficient (still takes ~15 sec on 1JJ2)
- upload page can deal with failed PDB/NDB upload in a basic way
- download from NDB works but is slow...
- connection to NDB FTP server takes ~30 seconds -- this is why so slow!
- added refinement program and temperature, R and Rfree to pdbstats page
- removed defunct images and mockup HTML pages
- added configuration page for Reduce -build
- incorporated Flipkin creation into the first Reduce step
- started porting the second, user-guided Reduce step
- finished the mechanics of porting the second Reduce step, now just UI tweaking
- updated the documentation on files created per model, and reorganized the sections
- added B-factor ribbons back to the multi-criteria kinemage
- made it so that second Reduce run is entirely skipped if no changes are requested
- marked models that have been through Reduce -build once; makes no sense to run them again!
- replaced progress bar image with a spinning ubiquitin ribbon drawing
- developed oneline-batch.php to extract basic quality stats from a folder of Reduced PDBs
- added calcCbetaStats() to find min, max, mode, mean, stddev
- improved REMARK regex's in pdbstat.php
- oops! I was calling the median "mode" in the above calcCbetaStats().
- code for tallying Rama outliers kept a running tally over all models (var not reset)
- Reduce 2.21 on Linux introduced line noise into some USER MODS, including ASCII 26 (hex 1a).
  Scrublines was set to terminate on this character, so I changed that and recompiled.
  This was previously causing affected PDBs to be truncated upon upload.
- improved regex's for R and Rfree to specify leading zero or decimal point
- changed loadClashlist() to guard against no clashes (and no summary!) for scores

### 3pre12 ###
- updated program requirements to list "zip"
- changed style sheet to sans-serif fonts, turned off link underlines
- changed color and upped contrast slightly on the logo
- added more elaborate flow chart
- created dummy pages for compare and improve
- created page to describe contents of uploaded file
- added Ramalyze PDF to analysis script
- swapped file upload and PDB-pull panes on upload page
- mc-mc dots in multicrit kin were actually all dots
- added reduceBuild() function to model.php
- implemented removeModel() and tested it from the command line scripts
- created a disk-too-full die() in mpInitEnvirons()
- added endsWith() to strings.php
- added system logging function. Individual user logging is of little use
  and will be effectively replaced by automated lab notebook entries.
- implemented a progress bar and list of tasks for analysis script
- did the same for addmodel and reduce-build jobs

### 3pre11 ###
- updated PDB stat to find nonpolar H in nucleic acids, too.
- updated pdbstat to find nucleic acid residues
- changed preparePDB() to not add H; now done by reduceNoBuild()
- renamed prepared() upload PDB so as not to confuse user
- reworked analysis scripts so that only selected jobs are performed
- reworked analysis so that hydrogens are added "just in time" if needed
- put both mc-mc and sc-(anything) dots in AAC kinemage, but separately
- removed Ramachandran JPEG and EPS files in anticipation of providing PDFs
- removed Mage as a necessary binary
- modified multicriterion kin to separate mc-mc and sc-x dots
- replaced findOutliers() with 4 separate functions
- analysis display page now adapts to the presence / absence of various criteria
- refactored jobs/addmodel.php into lib/model.php:addModel()
- refactored jobs/analyze.php into lib/analyze.php:runAnalysis()
- created a proof-of-concept cmd-line-driven script to use MP functions
- added machinery to create a ZIP archive of all a model's files

### 3pre10 ###
- really reworked page header/footer/styling. Style sheet hasn't been finished yet.
- removed all tab bar images and templates
- came up with a pretty decent "look" for the site; may change colors to match kinemage site
- rearranged upload page
- allowed user to revisit previously generated analysis
- each model now has its own download URL too
- implemented function to generate kinemage-viewing links
- added embedded KiNG page for kinemage viewing
- tried a pale brown background theme
- fixed accidentally introduced bug in analyze_display.php
- made nicer display of multi crit kinemage
- added Ramachandran plots to output of analysis (jpeg, kin, eps)
- added CB dev, all-atom contacts to analysis output

### 3pre9 ###
- added 'ignoreSegID' option to handle old Reduce'd files with 'new' in segID field
- loadRamachandran() and loadRotamer() now decompose residue names into parts
- loadCbetaDev() now synthesizes a residue name from the parts
- loadClashlist() now picks up worst clash per residue
- badRes now markes residues with clashes >= 0.40 A
- refactored badRes code into lib/analysis.php:findOutliers()
- started lib/visualize.php to make data visualizations
- wrote functions to find res with alt confs and guess at res center in 3-space
- started working on the new MultiCrit

### 3pre8 ###
- brought the whole project under Subversion version control
- created a shell script to restore filesystem state (symlinks and permissions)
- created analyze_display.php for Jeremy to start working in
- fixed loadCbetaDev() to deal with multiple headers from Prekin
- created a basic file-listing page
- administrator can now "enter" sessions from the monitoring page
- mocked up some HTML layouts for pages
- check script now tells about how big file uploads can be
- added "last touched" to admin page
- moved all string and formating functions to strings.php
- set svn:ignore properties on public_html/ and on symlinks in bin/
- check for presence of models before trying to list them.
  Alternately, we could just define $_SESSION['models'] = array() at startup.
  Actually, lets do that too.

### 3pre7 ###
- filled in the PDB upload form, though it doesn't do anything yet.
- implemented first cases of "More/Fewer options" links
- implemented background execution via launchBackground()
- implemented background job monitoring via job_progress.php
- file upload now works, basically, though no processing is performed.
- added config parameter MP_DEFAULT_TIMEZONE
- worked out a bug in launchBackground(): session data wasn't written immediately
- addmodel.php now does some nice pre-processing of all new models uploaded
- pdbstat now reports refinement program and temp, and a count of CNS headers
- refactored jobs/addmodel.php to facilitate future offline scripts
- created a function to write session data to disk immediately
- reworked the definition of a model entry.
  Now using -build will create a new model, so you can compare to the old.
- did proof-of-concept for Cbeta deviation analysis and clashlist

### 3pre6 ###
- filled in the Finish tab (rough draft) to allow users to log out
- described what should go in each tab

### 3pre5 ###
- fixed bug in GC function
- made PATH in config.php depend on MP_BASE_DIR
- added some things to check_config.php
- PATH setup is smarter: auto-detects OS and adds our bin/ directory
- added read-only marker for sessions to enable D&J's partial results
- it might make sense to make labbook keywords an array, but then
  transmitting them in a form becomes harder.
- started working on a standard page header
- Experimented with tables, images to make tabs. Images are a better bet.
- made first drafts of all the tab pages (empty). Linked them together.
- made labbook page work pretty well

### 3pre4 ###
- refined sessions yet again: now each top level page must know where
  the base installation directory is, relative to itself.
- session support has been reworked to use MP_BASE_DIR
- session create now defaults to false. This limits points of entry.
- added marker file for session lifetime
- added administrator overview of current sessions
- created garbage collection functions

### 3pre3 ###
- added (optional) mode to mkdir() call in sessions.php
- added time zone support to lab notebook
- continued refining session code

### 3pre2 ###
- wrote the which() part of check_config.php
- configured sessions to never use cookies
- implemented bare-bones of a lab notebook system
- rearranged how the include() system is set up

### 3pre1 ###
- built custom session-handling functions
- established simple test case for running background jobs
- started documentation