Thursday, January 8, 2009

Attendees: Elaine, Tom, Greg, Scooter, tef, Conrad

New Action Items

• Conrad will make sure that daily builds are installed on socrates2
• Conrad will generate deltas between 1.3 HEAD and 1.3 branch point so that everyone can verify whether all changes from 1.3 have been ported to 1.4
• Greg will change build script to compile OTF from svn tree instead of tar.gz file
• Eric will check on how chimera-bugs@… is used, both in conjunction with gnats and as manual bug report address
• Conrad will add two new agenda items: (1) what bugfixes should be back-ported to 1.3, and (2) update list of features to be removed.

Previous Action Items

• Greg will make production and candidate releases not have build numbers as part of the file name
• Greg will change the release string from "x.y.z.w" to "x.y.zbw" where the "b" is literal
• Greg will change daily build and snapshot naming to match the release string, ie include the build numer in the file names
• Scooter will convert gnats to trac
• Scooter and Eric create account for Wei to access svn/trac
• Conrad will look into disabling stereo in orthographic view
• Al will decommission SGI after the next Chimera production release
• tef will contact Adobe for more information about embedding 3d graphics in PDF documents

Minutes

1. Previous Action Items
   • Conversion from cvs to svn is complete
   • Stereo-capable workstation was installed as part of December open house

2. Tru64 builds (Conrad)
   • Compiling AMBER10 programs on Tru64 tickles a compiler bug. Since fixing this requires more effort than we are willing to expend on a deprecated platform, it was decided that we will no longer build on Tru64. As a replacement, Conrad will make sure that daily builds are installed on socrates2 so that a server-based version of Chimera is available to test remote display

3. svn conversion
   • Greg points out that svn does not handle merging after file reorganization on a branch very well. The solution is to reorganize only on the trunk. Since we do not do this often, it should not be much of a problem.
   • Tom asked whether we have checked that all changes to 1.3 have been ported to 1.4. Conrad will generate deltas between 1.3 HEAD and 1.3 branch point so that everyone can check their commits against 1.4. There should not be many, and merging individual files should be straightforward with svn.
   • As part of the cvs->svn switchover, OTF source was relocated to be part of the Chimera repository. Greg will change the Chimera build process to compile using the OTF source file instead of requiring a separate otf.tar.gz.
   • Tom pointed out that once you check out part of an svn tree, it is not possible to check out a single file afterwards. The workaround is to use a non-recursive checkout to get all the files (but not subdirectories) in a directory. It is recommended that developers pay the price and check out the entire repository once, and then use commits and updates to maintain an up-to-date copy of the source tree.

4. gnats->trac conversion

• Scooter reports that the e-mail interface works. Replies to a message sent by trac will be appended to the referenced issue without manual intervention.
• Scooter said that we need a flag day for converting from gnats to trac. As part of the change, the e-mail address used as the "From" (or Reply-to) address of trac messages need to be selected. Scooter suggested chimera-trac-bugs and Conrad asked why we cannot keep chimera-bugs. The issue is that we do not know whether gnats uses chimera-bugs as its From address. Replies to gnats are not handled by the trac scripts, so we do not want gnats replies to go through trac channels. Eric will check to what e-mail address are replies to gnats sent.
• Elaine points out that we need to update the Feedback web page on flag day.

5. Default view on opening molecule
   • Again, no decision (other than we want a smarter default) was reached.
   • Tom pointed out that there are actually three options when considering presets: apply to all models (ignoring what's selected), apply only to selected (ignoring unselected models), and apply to selected models but taking into account unselected ones. The third option was not mentioned previously and deserves further consideration.