Skip to content

[Bioc-devel] HAVE_ZLIB in affyio (breaking matchprobes)

3 messages · Ben Bolstad, Wolfgang Huber, Hervé Pagès

#
Wolfgang,

Sorry about that I'll fix up my part promptly. I do agree that
supporting compressed CEL files is a good idea and should remain
supported (though they have historically been used by only an extremely
small part of the userbase).

Ben
On Wed, 2006-04-05 at 18:17 +0100, Wolfgang Huber wrote:
#
Hi Ben, Herve,

can affyio be fixed such that the code that is conditionally compiled 
with #if defined(HAVE_ZLIB) preprocessor commands is indeed used if the 
machine does have zlib (which I think is standard on most modern systems).

Currently, this is breaking the build of matchprobes on the machines in 
Seattle:
http://www.bioconductor.org/checkResults/1.8/html/gopher5/matchprobes.buildsrc.html
since the example CEL files that the vignettes uses are gzipped, and 
affy::read.affybatch (via affyio/src/read_abatch.c) refuses to read it 
because it was apparently compiled without HAVE_ZLIB defined.

When I manually define HAVE_ZLIB in affyio/src/read_abatch.c, the 
matchprobes package passes R CMD check.

1. Can somebody (Ben?) add a configure.ac (or whatever helps) to affyio 
so that HAVE_ZLIB is defined?

2. Herve, can you make sure the build machines do install a version of 
affyio that supports compressed files?

Beyond fixing matchprobes, I think being able to read compressed CEL 
files is a good thing for everybody. Or am I missing sth?

Best regards
   Wolfgang

-------------------------------------
Wolfgang Huber
European Bioinformatics Institute
European Molecular Biology Laboratory
Cambridge CB10 1SD
England
Phone: +44 1223 494642
Fax:   +44 1223 494486
Http:  www.ebi.ac.uk/huber
#
Hi,

I just "R CMD build/check" matchprobes on gopher5 (our main build machine)
with the last affyio version that Ben committed to svn today (v 0.99.1).
Since everything went fine, I conclude that the new affyio was automatically
installed with support for compressed files (with no manual 
configuration required).
So expect to see some green for the matchprobes result in tomorrow's
build/check report...
Cheers,

H.
Ben Bolstad wrote: