[pony-build] Bug with force build

[C. Titus Brown]

On Fri, Apr 02, 2010 at 05:30:13PM -0400, Jack Carlson wrote:
> Meant to send this out yesterday, sorry for the delay.  Found the force-build issue a bit after our meeting yesterday (Thursday), but looks like it could be a bit of a pain to fix (or could even be ignorable).  In pony-client, this is how builds are checked if they should be run, and forced if they aren't requested:
> 
> if not options.force_build:
>     if not check(name, server_url, tags=tags):
>        [errormsg and exit]
> 
> So if a build is forced, check() isn't called, which means that the check_should_build() method of coordinator on the server side is never run.  check_should_build() seems to do more than simply check if building [not 100% sure what exactly, I suspect request.build.pop(tagset) though], which is a prerequisite for the server prepping to receive the build.  I'm not completely familiar with the server side of pony build, but it seems to me that check_should_build() shouldn't be doing that, but from a cursory glance at the issue looks like it could take a while to restructure various elements of the server side so that isn't needed (or is done somewhere else).
> 
> So unless I'm making this out to be a lot more difficult than it is, I figure I should just go ahead and continue on with the pony_client rework stuff instead of continuing with looking at fixing this issue, which could potentially negate the problem anyway--just have the reworked version call the check-if-should-build logic as something like:
> 
> if not check(...) and not force:
>   [errormsg and exit]
> 
> instead of how it currently is.  
> 
> Any thoughts?

First, I agree that this is probably the problem.  How would you to nail down
exactly what's happening?  (Hint: have you tried actually modifying any of
the problematic code?)

Second, could you post a bug report to my github issue tracker?

thanks!
--titus