[patch] to bug 82243 (not 88243)

[Germain Garand]

Le lundi 14 Février 2005 22:05, Paul Temple a écrit :
> Germain Garand wrote:
> > > Thanks! I must get testregression running...
> >
> > no problem, but I kind of expected you had... ;(
>
> It runs now, but I have many FAILS. So I'm not sure if it really
> works properly. I undid my last commit to xml/dom_nodeimpl.cpp,
> but that does not remove the FAILs.
>
> I post the first lines that are printed when I call
> ./testregression --noxvfb /blabla/kdecvs/khtmltests/regression
> Maybe someone can tell me what's wrong with it (I assume all
> tests should PASS).
>

it's not as straight forward as this...
testregression does its best to create an environment as reliable and stable 
as possible, but there might still be differences causing (mostly) [Paint] 
changes (such as font differences essentially) from baseline... 
here probably also the fact that you are using an ordinary X server, with 
different depth/extension/whatever (baseline was generated with Xvfb) will 
come into play.

Other differences are indeed temporary changes that have not been reviewed or 
not been checked in yet.

You can visually evaluate the changes by pointing your browser to 
khtmltests/regression/output/index.html afterward.

So: either you can match most of baseline, or, if that is definetly too far 
away, you may generate a local baseline _before_ any patch as follow:

cd /home/foo/khtmltests
cp -R regression regression_local
$KHTML/testregression -html -g  /home/foo/khtmltests/regression_local

then apply your patch, and check any difference against your baseline:

$KHTML/testregression -html /home/cvs/khtmltests/regression_local

Time consuming, but definitive.

Greetings,
Germain