[phobos] automated builds and tests

Benjamin Shropshire benjamin at precisionsoftware.us
Sun Sep 5 12:31:58 PDT 2010


Brad Roberts wrote:
> On 9/5/2010 7:23 AM, Benjamin Shropshire wrote:
>   
>> Brad Roberts wrote:
>>     
>>> http://d.puremagic.com/test-results/index.ghtml
>>>
>>> It's super crude.. it looks horrible.. but it's a start.
>>>
>>> What it does:
>>>    forever:
>>>      checkout a fresh copy of dmd, druntime, phobos
>>>      build dmd
>>>      build druntime
>>>      build phobos
>>>      test druntime
>>>      test phobos
>>>      test dmd (quick mode)
>>>      sleep 3600
>>>
>>> Anyone want to help make it not look like crap?
>>>
>>>   
>>>       
>> I've already started something similar. I'm planning on a super set of that
>> functionality: arbitrary number of tests, individually broken down, new test get
>> run on old versions of DMD, log per test compile times, run times and binary sizes.
>>
>> OTOH right now all I have is the DB schema and a few pages populated with fake
>> data.
>>     
>
> I'd be just as happy for someone else to own this, but I got sick of seeing
> regressions and decided to whip it up.  Only invested a few hours in it so far.
>
>   

What is your code in? Mines in Python, but a front-end/back-end language 
split wouldn't be a real problem.

> The failure in the dmd test suite is wierd.  I'm trying to figure out why the
> test in a clean checkout produces different results than in my development
> checkout.  There's darned few differences, but obviously something differs.
> It's creating the resulting .lst file in
> test_results/runnablerunnable-sieve.lst.  Easy enough to add an extra / in the
> dmd_coverDestPath call, but why is it acting different?  I think I'll 'fix' and
> then dig later.
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
>
>   



More information about the phobos mailing list