incubator-triplesoup-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From David Reid <>
Subject Re: testing stuff over HTTP
Date Tue, 20 Feb 2007 00:56:12 GMT
Garrett Rooney wrote:
> On 2/19/07, Leo Simons <> wrote:
>> On Feb 19, 2007, at 3:42 PM, David Reid wrote:
>> >> Obviously we want to use SPARQL for all the data fetching, but we
>> >> also
>> >> need to put the stuff in there to fetch, otherwise how can we test
>> >> properly? :-)
>> >
>> > Well, how about having an sqllite database (single file) that we can
>> > populate and then store in SVN?
>> That could work...if you look at what others have mostly done for
>> tests, they're either storing RDF/XML or RDF/N3...which I can
>> understand, since the human-readable-ness is nice for diffs and the
>> like...maybe storing the sqldump would get us the any case,
>> we should leech all those existing tests and use them if we can! :-)
> I would worry about using a sqlite db simply because changes to the
> tests would not show up as diffs in svn commits.  If you use a textual
> format (perhaps sucked into sqlite to run the tests) it's a lot more
> transparent and encourages more review of changes.

Good point.

Maybe we should aim for having the RDF data as RDF/turtle and also as
some form of SQL dump. Ideally the SQL dump could then be read by
multiple db's (not sure if this is even possible) but that would mean we
could test more aspects and should simplify the initial setting up.

>> > As for httpd, we could have a config file that set it up specifically
>> > for the tests, again stored in SVN. Part of the configure step
>> > could be
>> > to find the httpd executable (and check the version I guess). In fact
>> > mod_sparql currently uses apxs, so this will probably be required
>> > anyways.
> Are we just reinventing the Apache::Test wheel here?  The mod_perl
> guys have had this solved in a rock-solid way for quite some time.  I
> realize some people in the room are not the biggest fans of perl, but
> Apache:Test is really really slick...

If there's something out there that can simplify things and reduce the
amount of effort required I'm all for it :-) at the very least we need
to document what we need for the setup, so maybe starting a doc
explaining that would be a start?

View raw message