Re: RFC: avr-evtd
Hi, Charles.
Thank you very much for your reply.
On Sep 16 2008, Charles Plessy wrote:
> Le Tue, Sep 16, 2008 at 02:16:19AM -0300, Rogério Brito a écrit :
> > I would be glad if someone uploaded this package for me or provided
> > me with constructive criticism.
>
> Hi Rogério,
>
> I had a look to your package. I am not very familiar with that kind of
> software, so I will only make comments on the packaging.
It's a program for the Kurobox that you sent me. :-) Thanks you very
much once again.
> * Although Upstream does not seem to be active since two years, I would
> anyway recommend to publish your patches on SF.net. At least it could be
> useful to other distros, or encourage people to prepare a new release.
Right. I have already contacted upstream, but he is not that responsive.
:-( I offered help to getting things in a sourceforge repository and he
told me that he will make me a member of the project. I just don't know
how many privileges I will have there. :-(
> If you like the idea, then I suggest that once it is done you add the
> URL to the patch on SF.net in the header of the patch.
Nice.
> * Some of your patches fix coding style (whitespaces, DOS carriage
> returns, indentation). Consider that they may create a small work
> overhead to people who may have to investigate the package, for instance
> if there is a security issue while you are in vacation. They will also
> make it uneasy to exchange patches with Upstream or other distros when
> these patches are applied after the coding-style ones.
Right. I will produce another version of the package without them. I
really got mad because the coding is not that good (could be simplified)
and I have already stepped up to be upstream.
> * To fully comply with Policy 3.8.0, you can copy
> /usr/share/doc/quilt/README.source or mention it in
> debian/README.source.
Right, but I have one question: quilt/README.source just seems to be
teaching people how to work with quilt...
> * If you have free time, like the idea and want to help it gain
> momentum, you can consider converting your copyright file to the
> format detailed in http://wiki.debian.org/Proposals/CopyrightFormat.
> to the format.
Yes, I like this proposal and I have free time. Done.
> * Upstream chose -Os for compilation. -O2 in Debian is not strictly
> mandatory, so if you find that sticking to -Os helps, you can.
Yes, I agree.
I think that in this embedded system, each byte that we can save will be
a huge advantage to serve files. I'm looking for optimizing things for
throughput and compiling some packages with -Os is a good thing.
Actually, compiling things with -Os should be the recommended setting in
Debian, since less code often means that there are fewer L1-i cache
misses, and those misses can be quite costly. When things don't even fit
into the L2 cache, things get quite bad in terms of performance and this
is quite possibly the case with people running Desktops with zillions of
little processes.
I'm still conservative and am writing this with fluxbox and mutt. :-)
Call me old-fashioned.
Anyway, it's done for the package.
> * Upstream's makefile has some instructions (-DMIPS) if it detects mips
> CPU. You do not do the same test in debian/rules (an you do not use
> Upstream's makefile). Is it intentional?
Yes, it was intentional. I don't know if the method that upstream uses
is a good thing for Debian, and we will soon have mips and mipsel and
I'm unsure of how to proceed. Things work fine on powerpc. I think that
I will postpone this for now, but include the checks in a future release
of the package.
A new version was just uploaded to mentors.debian.net.
Thanks for your review, Rogério Brito.
--
Rogério Brito : rbrito@{mackenzie,ime.usp}.br : GPG key 1024D/7C2CAEB8
http://www.ime.usp.br/~rbrito : http://meusite.mackenzie.com.br/rbrito
Projects: algorithms.berlios.de : lame.sf.net : vrms.alioth.debian.org
Reply to: