On Oct 14, 2009, at 7:09 AM, kcrisman wrote:

>>> I don't think this would work at all, unless there is an *easy*
>>> automated procedure to submit a patch to a bot which (a) applies the
>>> patch to the current tree (b) compiles in all platforms (c) attaches
>>> the logs to the ticket and possibly flags the ticket depending on  
>>> the
>>> result of the test.
>>
>> A build bot would be nice, but I do not myself see the fact it is not
>> dead easy to test on another platform, and have a build bot
>> automatically attach a patch, should be an excuse for not checking. I
>> certainly check my patches on hardware other than what I develop it  
>> on.
>>
>
> No, it is very, very difficult for some of us.  Even for me to do the
> recent checking of PPC versus MacIntel I did to help fix that problem
> was a sheer quirk that my technology center hasn't reclaimed my old
> computer yet because I was having problems with my new one.  Soon I
> will have access to precisely one computer, let alone architecture.
> And, frankly, the time involved in checking (for me) on another
> computer is significantly greater than the time I just spent reading
> this thread :)

I'll second this. I try out my stuff on my own machine, and perhaps  
sage.math if I suspect it might be doing anything funky, but anything  
more is time I'd rather be spending doing something else. For someone  
who has a hobby of building complex software on a wide variety of  
systems (and I'm very glad people like you are out there as the Sage  
project wouldn't be where it is without it) doing this is no burden at  
all. I'd rather spend my time thinking about, e.g., algorithms than  
build processes.

My attitude is that (nearly) anything that can be automated should be  
automated. I get the impression that the job of release management  
involves a lot of acting as a manual build bot and waiting (stuff  
computers are good at) as well as conflict resolution. An automated  
build bot could almost completely manage the first task, and also help  
with the second (specifically, if testing were more granular, ideally  
on the patch rather than alpha level, there would be less conflict to  
resolve.

- Robert


--~--~---------~--~----~------------~-------~--~----~
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel-unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---

Reply via email to