[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: "Debian is switching to EGLIBC"

Aurelien Jarno wrote:
> Michael Prokop a écrit :
>> | Debian is switching to EGLIBC
>> |
>> | I have just uploaded Embedded GLIBC (EGLIBC) into the archive (it is
>> | currently waiting in the NEW queue), which will soon replace the GNU
>> | C Library (GLIBC).
>> | [...]
>>   -- http://blog.aurel32.net/?p=47
>> Where did this decission (and the discussion around it) took place?
>> I can't find anything neither on debian-devel nor on debian-devel-glibc.
> It has been discussed with a few people during Debconf, and on
> #debian-glibc and #debian-arm.

So I think the problem here is not that you made a technically bad
decision.  It sounds like you made a good decision. It's how it was

1) It didn't happen on any of the official Debian places that developers
read.  Nothing on debian-devel or d-d-a.

2) It didn't really convey how much of an impact this had.  How minor a
change it is -- that it is really a patch series atop glibc much in the
same manner as go-oo.org is to OpenOffice.

3) Announcing to small subgroups of developers (those that can afford to
to to conferences or that happen to be in an obscure IRC channel at the
right time of day) is no substitute for announcing to developers in general.

4) Posting on a blog is no substitute for posting to debian-devel.

glibc *is* a big deal.  I read your blog post and alarm bells went off
in my head (the project has a *goal* to maintain API/ABI compatibility,
but not a guarantee, for instance?)  The language about "switching" made
it worse.

I for one would have appreciated it if, before the upload, you had laid
out why you're planning to do it here on debian-devel.  I don't think
you would have met any opposition.

You have to realize that glibc is not just another package.  Where it
may not matter that much to the general public if openoffice.org
switches upstreams to a patch series, glibc is a different beast with
far longer tentacles in every bit of the system.

I, for one, have heard just about enough of "Hey developers, we're doing
$FOO, and it's already been decided, so put up or shut up" from people.
 I'd like a little bit more along the lines of "Hey developers, we
really think $FOO is a good idea.  Here's why.  What do you all think?"

-- John

Reply to: