Re: [WIP/RFH/PATCH 3/3] t/lib-git-svn: adjust config to apache 2.4

[Michael J Gruber <git@drmicha.warpmail.net>]

Jeff King venit, vidit, dixit 10.04.2015 05:39:
> On Fri, Apr 10, 2015 at 05:23:17AM +0200, Jeff King wrote:
> 
>> Curiously, though, the git-svn tests seem to run fine for me on Apache
>> 2.4 without your patch. I wonder if the fixes I mentioned above (which I
>> definitely needed to get the regular http tests running on Debian back
>> then) are not necessary on my system anymore (e.g., because the apache2
>> package now ships with better-compatible config).
> 
> Ah, I see. I am not in fact running apache.
> 
> If you do not set SVN_HTTPD_PORT (which I don't), then lib-git-svn.sh's
> start_httpd silently returns success without bothering to setup the
> apache server at all. And yet the rest of the tests run to completion
> just fine.

Yep.

> It looks like setting this variable is a way to say "run the same set of
> tests, but do it over svn-over-httpd instead of svn's direct filesystem
> access". Setting SVN_HTTPD_PORT does cause the tests to fail for me.

Oh, I didn't even notice that. That's a bit weird.

> I don't know how important it is to run these tests over httpd. If so,
> it would be nice to do something like lib-httpd does: try the apache
> setup by default, and if it fails, gracefully skip the tests (or
> fallback to the non-httpd version).
> 
> I'm also not sure if there's value in running both the httpd and
> local-filesystem forms of the test. IOW, should we be factoring out the
> tests and having two separate scripts that set $svnrepo as appropriate,
> and then runs the tests?
> 
> -Peff

Hmm, if those tests are run (with file://) I'm inclined to leave things
as they are (scratch 3/3)... Though the fact that on my system,
lib-git-svn starts the server but fails to connect isn't overly
comforting. But git-svn is being used less and less. World domination is
almost complete.

Michael