|
|
Boost Testing : |
From: Vladimir Prus (ghost_at_[hidden])
Date: 2006-02-22 02:40:03
Reece Dunn wrote:
> Martin Wille wrote:
>>Vladimir Prus wrote:
>>
>> > So, what is your opinion about switch to V2?
>>
>>I'm all for it.
>>
>>The procedure you suggest also sounds fine for me, except for the "next
>>day" part. A single full regression test run takes more than 24 hours
>>here.
>
> I also agree.
>
> NOTE: I am willing to get setup as a regression tester running V2. IIUC,
> this would be a case of running regression.py instead of the custom v1/v2
> script I am currently using.
>
> I have a few questions:
> * is it okay to upload the V2 results to the metacomm regression results
> area?
The plan is that as soon we start switching, the results will be uploaded
there.
> * where are the docs for setting up regression.py/getting started?
Preliminary version is:
https://zigzag.cs.msu.su/boost.build/wiki/BoostRegressionsV2
but we'd need to test it more before starting the switch.
> * is it possible to cycle through different toolsets, i.e. msvc one day,
> cw the next?
I think that by using different runner id, it's possible (say, Reece-msvc
and Reece-cw).
> * is it possible to run though the V1 tests with a given toolser and then
> the V2 tests with the same toolset and upload the results together, so you
> would have (for example) msvc/V1 and msvc/V2?
I think it's possible, by using different runner-id (Reece-V1 and Reece-V2)
> * is there a way to label the specific toolset results as V2/cw-9.2?
Again, runner id is the solution. I'll put up more specific instructions
when we're fully ready.
- Volodya