W3C home > Mailing lists > Public > public-css-testsuite@w3.org > March 2012

RE: [RC6] Several background-0* tests are duplicates

From: Arron Eicholz <Arron.Eicholz@microsoft.com>
Date: Mon, 26 Mar 2012 19:52:34 +0000
To: "css21testsuite@gtalbot.org" <css21testsuite@gtalbot.org>
CC: Public CSS test suite mailing list <public-css-testsuite@w3.org>
Message-ID: <07349ECFC3608F48BC3B10459913E70B1E4BD984@TK5EX14MBXC293.redmond.corp.microsoft.com>
On Monday, March  26, 2012 11:58 AM Gérard Talbot wrote:
> Le Lun 26 mars 2012 12:44, Arron Eicholz a écrit :
> > On Saturday, March 24, 2012 5:17 PM Gérard Talbot wrote:
> >> Le Sam 24 mars 2012 19:11, "Gérard Talbot" a écrit :
> >> > Hello,
> >> >
> >> > background-090 Background with (color image position repeat) [RC6]
> >> > http://test.csswg.org/suites/css2.1/20110323/html4/background-090.h
> >> > tm
> >> >
> >> > background-093 Background with (color repeat image position) [RC6]
> >> > http://test.csswg.org/suites/css2.1/20110323/html4/background-093.h
> >> > tm
> >> >
> >> > background-096 	Background with (color repeat position image)
> >> > [RC6]
> >> > http://test.csswg.org/suites/css2.1/20110323/html4/background-096.h
> >> > tm
> >> >
> >> > 'background'
> >> >     Value:  	[<'background-color'> || <'background-image'> ||
> >> > <'background-repeat'> || <'background-attachment'> ||
> >> > <'background-position'>] | inherit
> >> >
> >> > http://www.w3.org/TR/CSS21/colors.html#propdef-background
> >> >
> >> >
> >> >
> >> > "
> >> > A double bar (||) separates two or more options: one or more of
> >> > them must occur, in any order.
> >> > "
> >> > http://www.w3.org/TR/CSS21/about.html#value-defs
> >> >
> >> >
> >> > background-090, background-093 and background-096 have, for
> >> practical
> >> > reasons, the exact same code.
> >>
> >>
> >> background-111 Background with (image color repeat position) [RC6]
> >> http://test.csswg.org/suites/css2.1/20110323/html4/background-111.htm
> >>
> >> background-114 Background with (image color position repeat) [RC6]
> >> http://test.csswg.org/suites/css2.1/20110323/html4/background-114.htm
> >>
> >> background-117 Background with (image repeat color position) [RC6]
> >> http://test.csswg.org/suites/css2.1/20110323/html4/background-117.htm
> >>
> >> background-120 Background with (image repeat position color) [RC6]
> >> http://test.csswg.org/suites/css2.1/20110323/html4/background-120.htm
> >>
> >>
> >> Many tests are in reality duplicates; they have same code and same
> >> rendered layout, except permutation of background sub-properties.
> >>
> >
> > None of these that you list are duplicates. Visually they may look the
> > same but in reality they are testing that you can place the values in
> > any order as per how the specification defines. These tests are not
> > really relevant for established browsers since most of those browsers
> > have this logic correct but anyone implementing something new with a
> > new parser needs to run these cases and verify that the parsing is correct.
> > Also in the past (2+ years ago) multiple browsers had parsing errors
> > with many of the background tests. It wasn't until these tests were
> > written that all those bugs were caught.
> >
> > I do agree that there now may be a better way to handle parsing cases
> > for CSS but at the time these were written we only had visual
> > verification for parsing tests. If I were to rewrite these tests today
> > I would probably write a DOM test to verify that the values get parsed
> > correctly or if I were to leave them visual tests I would combine all
> > of them into one test and use smaller squares for verification. Either
> > way would work I think.
> >
> > Not sure what we should do now though. The tests are fine as is.
> 
> 
> The tests are indeed fine with me. I have created reftests for these.
> 
> 
> They
> > are valid, correct and they are not duplicates. So I am not seeing
> > your point on how they are duplicates.
> 
> 
> My limitations with english lead me to use "duplicates" word.
> 
> 
> > As long as the values are in a
> > different order and they do not have the same number of sub-values as
> > any other background test then it is a different test and not a
> > duplicate.
> 
> I see that other shorthand properties were tested in that manner too. E.g.
> 
> font-variant, font-weight, font-size and font-family
> http://test.csswg.org/suites/css2.1/20110323/html4/font-008.htm
> 
> font-weight, font-variant, font-size and font-family
> http://test.csswg.org/suites/css2.1/20110323/html4/font-010.htm
> 

Yes I think most of the shorthand properties were tested in this same manner. I think we should add some documentation on how we want to cover these scenarios in the future. I hope that we can have the same type of coverage for shorthands just tested in a more compact test case that handles all the combinations.

--
Thanks,
Arron Eicholz
Received on Monday, 26 March 2012 19:53:31 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 26 March 2012 19:53:39 GMT