EMBOSS: Project Meeting (Jan 31st 2000)


Attendees

Sanger Centre: Peter Rice, Richard Bruskiewich, Simon Kelley
HGMP: Alan Bleasby, Gary Williams, Val Curwen, Sinead O'Leary, Mark Faller
EBI:
Apologies: Ian Longden

1. Matters Arising

2. General progress on release 0.0.4

Graphics output device settings are fixed. There is a remaining problem with file naming for multiple graphics output files which is an internal PLPLOT library bug/feature.

Supermatcher is fixed. Peter is modifying ACD files to use "info" as the standard prompt and help attribute, with "prompt" and "help" only when they are different, to make writing of parsers (e.g. for PISE) simpler.

PISE developers have reported problems with HTML markup in some automated ACD output. Alternatives are under investigation.

Sets of Sets (SOS) data type still has to be implemented.

TOPO colour options are reimplemented as text codes.

entrails, a program to dump out internal data definitions for use in parsers, is still at an early stage of development.

Ian and Peter will work on feature input and processing. This includes checking for missing hooks in the code for processing a sequence with features (as in the "make check" application "seqretfeat") with all possible USAs. Alternative feature format input will be provided (EMBL, XML) and feature processing functions wil be updated for iteration, sorting, etc.

Peter may have a practical solution to generating standalone HTML documentation from the existing application web pages, by implementing a standard header that points to the pages currently listed in the sidebar.

Gary suggested inocluding the source HTML pages in the CVS tree. Peter wil do this, and include the script that generates skeleton pages for new applications.

Richard suggested looking into DocBook for documentation. At Sanger, ACEDB and ENSEMBL are considering DocBook so there would be a set of utilities available.

Peter has fixed a problem in processing multiple accession numbers for EMBL database entries. Val reported a problem with accession numbers that appear in more than one entry, as primary and secondary, in applications such as "seqinfo" which can hang. This may be part of the same problem. Val will check whether it still hangs.

Sylvia Nagl (University College London) is teaching an EMBOSS course this week using Val's tutorial.

3. Windows build

Simon attended to join the discussion on building EMBOSS for Windows. The current build is under CygWin on Windows NT and has been copied to Windows 95 by Richard where it appears to run without obvious problems.

CygWin seems to accept POSIX (Unix-style) file paths and names, although problems are reported with ":" in filenames. As EMBOSS usually strips these in USA processing this should be no problem. Backslash appears to be accepted in filenames too.

Any "system" calls will fail without CygWin due to the absence of the Unix shell. CygWin provides SH.EXE which can be copied with the distribution and defined by SH_PATH. For EMBOSS it is expected that the ajSystem call, which calls exec functions directly, will avoid this problem although a few calls may need to be replaced.

The executables built with static linking to the libraries are too large (25Mbytes). Dynamic linking should save space. This needs a list of externally available functions for each library. Peter will generate this list from SRS or from the documentation processing script which builds the EFUNC data files for SRS.

The ACD directory needs to be specified on installation. The EMBOSS_ACDROOT variable should do this. Checks are needed for whether other files and directories are successfully set or whether additional code/variables will be needed.

For Xwindows output, there is an "rxvt" xterm lookalike which uses an xlib API to make Win-32 calls. This could be linked to PLPLOT to produce Xwindows output on Windows systems.

4. Any Other Business

5. Next meeting

Next meeting will be Monday 7th February 2000, 11:00am, usual place.