Subversion repository and wiki

[Bjoern Hoehrmann]

* Cameron McCormack wrote:
>I realise that nothing much concrete has been discussed regarding how
>the tests should be structured, but I think it would be good to get
>things started and add some simple tests in there, and then we can see
>what sort of functionality we need.  Don't worry if changes need to be
>made later, since it's early on and won't be much trouble.

I've added 19 tests so we can work on fleshing this stuff out. I had the
following problems:

  * Putting each test in its own directory does not seem very nice,
    most tests are just a single file so we have lots of 1 file
    directories

  * Coming up with names for these directories is hard, I've tried to
    adopt the conventions used in all the other tests, but as there was
    just one... Working with prefixes doesn't work out very well, if
    you have a test that combines DOM, CSS, SMIL features, you'd either
    have lots of prefixes or you can't get all CSS tests simply by
    looking at the prefixes

  * some of the tests aren't backed up 200% by the specifications,
    I guess we are all painfully aware that SVG 1.1 has many bugs
    the SVG WG does not attempt to address; I'm not sure what to
    do about those; I've not marked the tests which might be con-
    troversial

  * it's not clear whether we should only have conformance tests or
    also, say, performance tests. I've added one, perf-typewriter-1,
    that almost kills Opera9 while it's about acceptable in ASV6.
    I have other tests that work e.g. in Opera9 to some acceptable
    level but almost kill Batik. Maybe we should have a separate
    top level directory for those...

  * The tests don't have any meta data other than the directory
    name. I'm happy with that, much meta data (like, depends on
    scripting, animation, etc.) can be automatically extracted from
    the tests, I don't think there is need for much more; I'd in
    fact rather get rid of the names.

  * The tests are probably inconsistent with respect to whether
    they use green or red when a specific feature is not supported,
    e.g. <svg><rect fill=red><script>rect.setFill(green)... vs
    <svg><rect fill=green><script>...rect.setFill(red)... if the
    viewer does not support scritping, you'll get inconsistent
    results; we might need a guideline for that

  * use-base-uri-1 has one start SVG document and the other is just
    a resource document. There would need to be some way to indicate
    that only one of them is significant

  * svn.svgtest.org does not seem to be configured to serve .svg
    documents using image/svg+xml so it seems they are not that
    accessible...

  * There are style sheet issues with the Wiki and I could not
    figure out how to edit pages... maybe there is some permission
    bit not flipped

The tests are also a bit outdated, but it's probably easier to work out
the test suite documentation from fixing the bugs I've introduced... :-)
-- 
Björn Höhrmann · mailto:bjoern at hoehrmann.de · http://bjoern.hoehrmann.de
Weinh. Str. 22 · Telefon: +49(0)621/4309674 · http://www.bjoernsworld.de
68309 Mannheim · PGP Pub. KeyID: 0xA4357E78 · http://www.websitedev.de/