[darcs-users] Re: Executing a program when applying patches

Mark Stosberg mark at summersault.com
Mon Nov 8 13:05:19 UTC 2004

On 2004-11-08, David Roundy <droundy at abridgegame.org> wrote:
>> I think this option for 'apply' could help in this case: 
>>        --leave-test-directory  don't remove the test directory
> No, this option doesn't allow you to reuse the test directory, it just
> leaves it there so you can examine it afterwards.  This can be useful if
> you want to see what happened when a test failed.

I agree the current use is ...useful, but having a persistent test
directory could be useful, too. As long as the tree is only used for
testing and remains "pristine", it should work well. The one hangup
I see is with old auto-generated files interfering with new ones, but
that could be a problem with the code. 

> For large projects, the time spent copying the files I would expect to be
> insignificant when compared with the time spend running the test.  I
> suppose there could be quick tests, but most commonly one will want to
> compile all the many many files in the large repository, which will take
> much longer than just copying them does.

You assume that files being managed need to be compiled to be tested. :) 
My biggest project I want to manage is in Perl. Our test suite is a
little different: Because the code comprises a website, our tests use 
a Perl-based web browser to 'visit' a copy of the website and test how
the code runs under Apache/mod_perl.

Also, there are some 10 MB movies setting in the tree, which can take
some time to copy. 

But OK, yes, it still takes a lot longer to run the tests than to copy
the tree. 



More information about the darcs-users mailing list