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

Re: Updates lacking descriptions

Hash: SHA1

Michael Schwendt wrote:

> On Thu, 13 Aug 2009 15:51:57 +0200, Kevin wrote:
>> AIUI, the package changelog only really needs to contain what 
you changed in 
>> the specfile,
> Tell that all the package maintainers, who do it differently.

I think he was referring to a minimum ("only really needs"). If 
you do more detailed changelogs, more power to you. I however 
have changelog entries just because a BR was missing or such. I 
don't see the need to add feature changes to the spec file which 
sees the software as nothing more than a blob to be extracted 
and commands run on. Another way: a hammer (spec file) doesn't 
care what the nail (software) is used for, only that it's a 

>> > Overall, however, what updates need is feedback from actual 
testers before
>> > they are marked stable.
>> That's really a separate issue from the lack of details.
> For me it isn't.  I won't spend extra time on writing special 
> for a test-update, if nobody contributes any testing. Once the 
update has
> been marked stable, it's too late. It will be installed by 
some users
> "blindly". They won't base any decision on reading the update
> description. Not even the list of affected bugzilla tickets in 
> implies that the fixes are correct or won't cause side-
effects. A
> minor feature addition might cause the software to crash in 
> environments. The update description doesn't add any quality. 
It just
> adds some eye-candy and creates another place where the 
packager can
> fail (and miss details or include wrong information, for 
> Packaging quality depends on other factors, such as the amount
> of time a package maintainer spends on cherry-picking upstream 
> or snapshots, self-contributed bug-fixes and testing prior to 

Maybe an updates-testing report (like the daily Rawhide Report 
which I scan even though I don't run Rawhide yet) could do some 

> Same applies to online changelog URLs. Tracking down locations 
on upstream
> websites (and verifying them with every update) is nothing 
else than an
> unnecessary burden. Just as upstream authors don't maintain 
> tarball ChangeLogs always, there is no guarantee that an 
> changelog would be up-to-date.

I have a folder full of bookmarks for important sites for the 
packages I maintain. Works well enough. I imagine it would scale 
well too.

- --Ben
Version: GnuPG v1.4.9 (GNU/Linux)


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