Re: Debian bug tracking system contains patch for GPM C-z hang
Hrvoje Niksic <hniksic(a)srce.hr> writes:
Problem #2: a while ago, they "froze" Debian with Wget
1.4.4.
Just so you know: They just froze again.
However, I soon released a bugfix 1.4.5, which fixed two or perhaps
three 1.4.5 bugs. They never bothered to add 1.4.5 to the
distribution, because the only updates were [sic!] bugfix-only.
This is weird. The freeze is there to make sure ONLY bugfixes come in.
Maybe the problem is that bugfixes in this case seems to be defined
more narrowly as "fixes for things as reported in the Debian bug
databsse".
(Note that [possible due to your complaints] their stable version does
contain 1.4.5. The frozen version contains 1.5.3. The recent SUSE 5.3
for instance ships with 1.5.2)
> Does it bother anyone, apart from me, that that here psfrag 3.0
is
> disabled in favor of 2.0 in teTeX although the psfrag authors would
> like very much to see it die, etc etc?
If I were psfrag author, it would bother the hell out of me.
So it does for me (it broke more than half of our exercise sets), so I
reverted it back again in our group setup. Thank god for symlinks.
However it still another case where there is a difference of opinion
about a design decision between the package author and our
users/sysadmins. In this case our sysadmins decided (rightly or wrongly)
to change this in our local version (in this case by reverting to the
old version). I think that the world is full of these local hacks by
sysadmins and that phenomenon of Linux distributions is just
increasing this to another level.
Yup. I believe there should be a closer relation between author and
distribution maintainer of a program, especially when the author is
willing to cooperate.
Yes. In particular the distribution-freeze method has severe problems
when the author also does freeze cycles. For instance Debian 2.1 will
still ship with 20.4. In this case this is not such a problem as 21.0
is too major a change to just follow. But imagine if this were a
21.0-21.1 thing and we release 21.1 just after the freeze. What should
be happening is that the distribution is tracking our bug fix betas in
their unstable branch, i.e. their freeze equals our freeze.
Also the term "bugfix" is very vague: The reading of stdin I
reimplemented in gnuclient recently, was that a new feature of a bug
fix?
I have long been intending to write something up for this to sent to
Debian (and maybe others) "officially", but have not got arround to
this.
Jan