preliminary packaging work of 6.4.5

[Thibaut VARÈNE]

Le 22 févr. 09 à 19:45, Rene Mayorga a écrit :

> On Sun, Feb 22, 2009 at 04:30:27AM +0100, Thibaut VARÈNE wrote:
>> I'd rather do it the good old way, tarballs 'n patches. You're  
>> welcome
>> to commit my changes wherever/however you see fit though.
>>
> Ok, I will work today on this.

Cool thanks! I need to figure an easy way for me to stay in sync with  
what you're working on (a way to stay in sync with the content of your  
current debian/ folder, that is ;)

>> I took a quick look at debian/changelog through the git web  
>> interface,
>> and I'm not sure I quite understand the purpose of
>> 005_disable_client_res.patch. If it's supposed to fix the build  
>> failure,
>> I think the better fix is to move the png icons where they were  
>> supposed
>> to be in the first place, as I do in rules...
>
> Good, at this momment I just did a rush to the package to build, and
> don't get on the details, I just stopped on debian/copyright and  
> then I
> was planning to move to all the others details,
> now that you already start with this the best is to just integreta the
> changes.

Indeed. :)

>> Regarding making BOINC point to boinc in /usr/include, I don't think
>> it's a good idea: people should update their code to the proper
>> location. Helping them to not have to do so is not really helping  
>> them
>> in the long term, I believe. That's my 2c :) Plus, let's not forget  
>> we
>> have only 2 known users of this code: boinc-app-seti and boinc-app-
>> milkyway. I've already updated the latter ;)
>
> I was planning to leave the symlink for a couple of release, but
> definitely try to not push that symlink to squeeze.

OK. My take on this is "the sooner the better" ;)

>> I see you've already updated debian/copyright, that's cool!
>
> Now there is the need to take a decision about some files:
> as I pointed out on debian/TODO.Debian:

yup. Now here's where I have to come straight I think: when I said  
earlier I wouldn't step up as a maintainer, that's because I have very  
limited free time, and barely enough to handle my own packages. It so  
happens that working on BOINC at the moment is aligned with my own  
(professional) interests, so I can help you to the extent these  
interests remain so aligned with your needs. As a matter of fact (and  
I know this will sound lame) dealing with copyright and licensing  
issues is quite out of the scope of my investigations. I do not plan  
to distribute what I'm doing, and as long as Berkeley is fine with  
distributing some source files, I'm fine with using them. Whether  
Debian as an entity sees things differently - which I can perfectly  
understand - it's none of my concern, as a "non-maintainer" of that  
package. ;P

Put otherwise, I can share with you whatever I do to make the package  
work (as in build, install and run fine), but that's about it.  
Strictly *packaging* issues (such as file licensing) are out of my  
current scope. I hope you understand that and I hope that the extent  
of the help I can provide you with for now will be enough for you.  
Plus, it's likely you won't be interested in some of my changes (such  
as keeping upstream hardcoded optimization flags, or using the CUDA  
libraries, as I do).

Speaking of which, if you're looking into distributability, as I  
pointed earlier, the coprocs/ folder contains the Nvidia CUDA binary  
libraries (for GPU computing, used by e.g. GPUGrid, SETI and  
Einstein), and I'd be very surprised if Debian would allow them to end  
up in a tarball on their servers. At least not in "main" (maybe non- 
free?) ;^P It's likely export-boinc.sh will need one more "rm -rf"  
line :)

>> With little
>> luck, merging both our works we might end up with something worth
>> releasing :) The only remaining red flag I see is the new default  
>> quit
>> message of boincmgr.
>
> ack.

I'll take a dive at the code soon. If I can see an easy way to get rid  
of this message, I'll add a patch to debian/patches.

>> I'll try merging this in the next couple days, will keep you posted.
>> Also, please *do* CC-me on replies: I'm not (yet) subscribed to the
>> mailing list, I saw your reply through the web interface... :P
>
> that is wear, I remember that you was CC-ed on the previous mail, hope
> this one make it to your mailbox or I'll have to blame my MTA :)

Actually, blame my MX: your email finally reached me after about a  
day ;)

> I'll start merging your changes on git today, if I did not finish or  
> if
> there is any further changes you can just send patches and I'll happy
> merge those on git :)


Excellent, that sounds perfect to me :)

Cheers

-- 
Thibaut VARÈNE
http://www.parisc-linux.org/~varenet/