xmlgraphics-batik-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Cameron McCormack <...@mcc.id.au>
Subject Re: Regard: setup and results
Date Wed, 08 Apr 2009 01:47:27 GMT
Hi Helder.

Helder Magalhães:
> I've been struggling with "regard" for a while so I wanted to share a
> few thoughts... :-)

Yes, the regression test suite does need a bit of love.

> 1. About the setup, it seems that the information is sparse and
> incomplete. I've noticed that the online documentation [1] misses
> setup steps, which I've found in a mailing list thread [2] and in the
> "MAINTAIN" file. I'm proposing the following:
>  * Add a "setup" section to the online documentation [1];

Sounds good.

>  * Move the running regard instructions within the "MAINTAIN" file a
> bit more towards the top. Running regard is in step 7 and is stated
> that, if it fails, one should go back to step 1. Wasn't it more useful
> if regard was one of the first actions?

Indeed. :)

>  * In the setup instructions, add a direct link to BE test suite
> archive [3], as the instructions in "MAINTAIN" file point to a page
> [4] where the information seems no longer available (not even in the
> more specific Test Suite Overview page [5]). Should this be pointed
> out to the SVG WG instead (Cameron)?

Including the direct link would be good.  The BE test suite is quite
old, in fact.  There have been numerous updates to the test suite
published by the SVG WG since then.  It would be great if we could get
the latest SVG 1.1 test suite run in regard, rather than the BE test

> Should I try to marsh up a patch containing this? If so, should I
> attach it here or create a report in the bug tracker to handle this?

Create a bug, please.

> 2. About a few runs results, I've noticed a few relevant rendering regressions:
>  * samples/tests/spec/paints/linearGradientLine.svg (missing most
> relevant part of rendering);
>  * samples/tests/spec/paints/radialGradientLine.svg (missing most
> relevant part of rendering);
>  * samples/tests/spec/paints/gradientPoint.svg (missing most relevant
> part of rendering).

I think that those tests are invalid, since a clarification was made to
the spec and official test suite about how objectBoundingBox gradients
should work when the the bounding box has zero width or height.

(I haven’t got a pointer right now to where the clarification was, but I
remember committing the change.)

> Also, minor potential color matching regressions or JVM improvements
> (not being familiar with color matching causes me not to tell the
> difference):
>  * samples/tests/spec/color/colorProfile.svg (colors are quite different);
>  * samples/tests/spec/scripting/paintType.svg (too much red? on the
> rectangle at position [2,2]);
>  * samples/tests/spec/masking/maskRegions.svg (too much red? on the
> masked shapes);
>  * color-colorProf-BE-03.svg (colors are quite different in the top
> right group of regtangles);
>  * RHints-RenderingCheck (too much red/too much green on the pattern
> using bicubic resampling)
>  * RHints-ContextRenderingCheck (too much red/too much green on the
> pattern using bicubic resampling)

I guess these will need careful analysis to determine if the resulting
colours are correct.

> Batik revision used in this test run was 761228 (the most up-to-date
> as of now). I'm on Windows XP SP3 and using JDK 1.6 Update 12. The JDK
> version could itself be a source of some regressions, but I'm
> currently experiencing a problem with JDK 1.4 update 19:
> Running jarCheckLoadSameAsDocument(scripts=application/java-archive)(scriptOrigin=any)(secure=true)
> java.lang.UnsupportedClassVersionError:
> org/apache/batik/bridge/IWasLoadedToo (Unsupported major.minor version
> 50.0)

Did you clean the build before running the tests under 1.4?  Perhaps the
IWasLoadedToo class was built while you were testing 1.6.

> So I've only checked that loading
> "samples/tests/spec/paints/linearGradientLine.svg" in Squiggle
> (build/run by JDK 1.4 update 19) also shows the same rendering result.

I think the regard test suite needs updating for these zero size
bounding box gradient tests.

Cameron McCormack ≝ http://mcc.id.au/

To unsubscribe, e-mail: batik-dev-unsubscribe@xmlgraphics.apache.org
For additional commands, e-mail: batik-dev-help@xmlgraphics.apache.org

View raw message