Please fix your bugs in release instead of rawhide!

[Joe Nall]

On Mar 10, 2009, at 9:48 AM, Jeremy Katz wrote:

> On Tuesday, March 10 2009, Christoph Wickert said:
>> Am Dienstag, den 10.03.2009, 15:21 +0530 schrieb Rahul Sundaram:
>>> Christoph Wickert wrote:
>>>> A note to all Fedora maintainers and especially to the Red Hat  
>>>> people
>>>> who are involved in development every day:
>>>>
>>>>           Please fix your bugs [1] in the release they were
>>>>         filed against instead of just closing them NEXTRELASE!
>>>
>>> Doesn't that depend on how serious the bugs are? Sometimes  
>>> upstream has
>>> gone way ahead, changed their configuration format etc and  
>>> backporting
>>> the fix or just pushing the update with all the other changes is too
>>> invasive.
>>
>> Agreed, but in the case I have in mind
>>      * the bug is so serious that he makes the package not working
>>      * we are upstream and
>>      * there is nothing to backport. Nevertheless the maintainer told
>>        me to get the rawhide package.
>
> So, as the person who has apparently pissed you off this morning, it's
> not as trivial as you're making it out to be.  I'm sorry that you want
> to be doing the development of the live images for the next release on
> Fedora 10, but the fact of the matter is that as the base system of
> Fedora moves onward, it's not just a foregone conclusion that  
> everything
> can be done on the previous release.  Yes, I could push back the  
> single
> patch for the syslinux path change... but that *isn't* sufficient to
> make a working rawhide live image while on Fedora 10.  You also need
> things like a) a newer kernel so that your system doesn't livelock  
> while
> writing to ext4 b) a newer squashfs-tools so that you can build an  
> image
> that works with the rawhide kernel[1] c) one or two other things are a
> little different and could have side effects -- sure, those could be
> tested extensively and dealt with/fixed/whatnot.  But that's then  
> effort
> not spent on implementing things for the future.  And given the  
> intended
> audience of livecd-tools, that's really not the best use of effort.
>
> Jeremy
>
> [1] And this one is even more insidious because the new mksquashfs  
> won't
> build an image that works with old kernels... so we push it back to  
> F10
> and all of sudden, building F10 images breaks for people.

Is the squashfs change the holdup on getting 2.6.29 in F10? Last time  
I tried the koji 2.6.29 F10 kernel, this was an issue in doing a  
custom application respin.

joe