tests: the cd-level issue

Mads Kiilerich mads at kiilerich.com
Tue Jun 12 05:56:25 CDT 2012


On 12/06/12 10:27, Pierre-Yves David wrote:
> On Tue, Jun 12, 2012 at 09:09:24AM +0200, Adrian Buehlmann wrote:
>>> Running a quick test with a local hack 'on demand' is IMO good enough
>>> for something like this. You are welcome to clean it up and make it
>>> cross platform ... and make a control mechanism or a commit description
>>> that justify that it is worth running all the time.
>> With regards to the aspect of "worth running all the time":
>>
>> How about adding an option to run-tests.py which is off by default? So
>> that the price for that check wouldn't have to be paid each time the
>> testsuite is run?
>>
>> Perhaps (run-tests.py --help):
>>
>>    --cdcheck             validate cd commands
>>
>> Not everyone is that good a doing local hacks on demand. Also, If you
>> alreday have cooked a local hack, it could perhaps be preserved into
>> run-tests.py.
> I do not really think that the "cost" of checking this worth an additional
> option. I expect such check to be very cheap comparing to everything else we do
> during tests.
>
> Do you have real number showing a terrible impact on the test suite ?

No, I don't have any numbers. It is very possible that more time has 
been spent discussing it than ever would be spent running it.

If you think it is worth pursuing then please go ahead.

/Mads



More information about the Mercurial-devel mailing list