[Bioperl-l] PPM4 PPD + server configuration

[Nathan S. Haigh]

Chris Fields wrote:
>   
>> -----Original Message-----
>> From: bioperl-l-bounces at lists.open-bio.org [mailto:bioperl-l-
>> bounces at lists.open-bio.org] On Behalf Of Nathan S. Haigh
>> Sent: Tuesday, September 26, 2006 4:17 PM
>> To: Bioperl-l at lists.open-bio.org
>> Subject: [Bioperl-l] PPM4 PPD + server configuration
>>
>> I've now have working packages for Bioperl 5.6.2-c, and Bundle::BioPerl
>> 2.6.(if it's decided that we need it) and a few other modules
>> (Bio-ASN1-EntrezGene 1.09, Class-AutoClass 1.01, GD-SVG 0.28 and
>> SVG-Graph 0.01). Please comment on the Bioperl and Bundle::BioPerl
>> packages with respect to thoe modules I've included as being "core" to
>> Bioperl, and those I've put into Bundle::bioPerl - relevant PPD XML code
>> at the bottom. Once, I've had feedback I'll also update the PREREQ_PM in
>> Makefile.PL.
>>     
>
> I think you mean 'Bioperl 1.5.2-c', right?  I wouldn't package a PPM for
> 1.5.2-c (which was a quick testing tag).  I think the idea is to get a
> 1.5.2-RC1 out next week, a possible second RC soon after, then the final
> 1.5.2 release, so you could make PPMs for those releases (using the same ppd
> file reqs and Bundle-Bioperl PPM).  
>   
Yes i did mean 1.5.2-c! I was too tiered by then! :o( For my own sanity,
I gave them the '-c' suffix. I'll be away next week, but someone should
be more than capable of modifying the versions in the package.xml file
and generating the .tar.gz files required :o)
>   
>> During this time, I stumbled across a bug in PPM4:
>> http://bugs.activestate.com/show_bug.cgi?id=50728. In short, PPM4 tries
>> to get some of the dependencies that are actually in ActiveStates repo
>> from the repo that the package is being installed from. For example, if
>> You were to try and install Bioperl 1.5.2 from the bioperl repo
>> (http://bioperl.org/DIST/) it will also try to get some dependencies
>> from this repo (which aren't there) instead of ActiveStates repo. A
>> workaround is to set up a redirect in the servers httpd.conf file,
>> something like the following should work for the Bioperl server:
>>
>> RedirectMatch /DIST/MSWin32-x86-multi-thread-5.8/(.*)
>> http://ppm.activestate.com/PPMPackages/5.8-windows/MSWin32-x86-multi-
>> thread-5.8/$1
>>
>> The bug has been fixed since I wrote the draft of this e-mail and will
>> be fixed in ActivePerl 5.8.8.820. However, they tell me, this won't be
>> released for at least a few months. Therefore, we need to implement the
>> above workaround until then - a volunteer?
>>
>> Cheers
>> Nath
>>     
>  
> We may have to get Mauricio or Chris D. on the httpd.conf addition; I don't
> think I have access to the server.  
>
> The Ace package is listed in the ActiveState repository as AcePerl, I think.
> Are the REQUIRE_NAME tags which contain the modules supposed to end with the
> double colon now?  I noticed "DB_File::" and "BerkeleyDB::" in the below.
>   
Thats right, PPM4 prefers to use the PROVIDE and REQUIRE tags where you
use the module name. If the module name doesn't explicitly contain a
double colon it is appended to the end of the module name. PPM4 uses the
PROVIDE tag to identify which package will install the required
dependency. So in the case of Ace, the AcePerl Package will contain
<PROVIDE NAME="Ace::">. Strictly speaking, the Bioperl PPD should
include PROVIDE tags for all it's modules! Much like the old Bioperl PPD
that is automatically generated in the ActiveStates repository. This is
why using %packages in the PREREQ_PM in Makefile.PL will now be better
and easier to maintain for creating PPDs.

> So, the intent is to install everything to get Bioperl running?  I wasn't
> sure about adding Bundle::BioPerl as a requirement, but I can see why, at
> least for Windows users.
>
> Chris
>   
I've done it this way, so that it's easy to remove Bundle::BioPerl if we
decide  not to do it this way, or, simply copy+paste the BioPerl REQUIRE
tags over to Bioperl if we decide that Bundle::BioPerl will be deprecated.

Cheers
Nath