Re: Opinion polls UIKit and Website...

[David Chisnall]

On 20 Dec 2013, at 11:34, Graham Lee <graham@iamleeg.com> wrote:

> That assertion is not demonstrated in practice. There are bugs with patches 
> attached in Savannah going back years, and the patches section has something 
> still there from 2004. Saying “we want patches” is one thing, but if they’re 
> not accepted then contributors will feel unwelcome and will move on.

$ svn log -r31303:HEAD http://svn.gna.org/svn/gnustep/libs/ | grep -i 'patch 
by'  | wc -l
      93

So at least 93 patches have been accepted in the last year.  Some committers 
forget to put 'Patch by', or put 'submitted by' or something else in the log, 
so this is not complete, but it definitely refutes the assertion that patches 
are not accepted.  Looking at the runtime, I've committed 22 patches by other 
people, and the runtime is one of the smaller projects in the GNUstep umbrella.

The bug tracker is an embarrassment.  There is no connection between GNA 
accounts (for svn) and Savannah accounts (for the bug tracker).  I tried to 
create a Savannah account twice and have failed both times.  Even if you do, 
the Savannah bug tracker is quite painful to work with (and I'm used to 
Bugzilla, so my standards are not very high when it comes to bug tracking).  
And some people decide to file bugs on GitHub, because we also have a bug 
tracker there[1], but most committers are completely unaware of it, while most 
non-cmmitters are unaware of the Savannah one (and why would they know about 
it?  The official site for GNUstep is gna.org, so why would you look on 
savannah.gnu.org for a bug tracker?).  

The only times I ever see bug reports for GNUstep is when Fred emails me.  
There's no automatic stuff, and no one is responsible for bug triage.  This is 
why I always recommend GNUstep contributors send patches to the mailing list, 
not via Savannah: patches sent to the list are reviewed and committed quickly.

In other projects I contribute to, each category in the bug tracker has a 
mailing list associated with it, and so people get a mail when there is a new 
bug and can assign it correctly.  In GNUstep, it just gets hidden away until 
someone is bored enough to poll the bug tracker.  

David

[1] https://github.com/gnustep/gnustep-base/issues

-- Sent from my Apple II