- From: Tantek Çelik <tantek@cs.stanford.edu>
- Date: Sat, 13 Mar 2004 10:57:11 -0800
- To: Ian Hickson <ian@hixie.ch>, "L. David Baron" <dbaron@dbaron.org>
- Cc: <public-css-testsuite@w3.org>
I concur with Ian's comments have only two other quick short comments to make: 1. serve the files as text/html for ease of trying them out in more browsers (alternatively, conditionally serve as application/xhtml+xml only if http accept headers allow for it -- folks have documented this on the net.) I know, this violates what is allowed for XHTML 1.1, but it helps usability (a good tradeoff), and will only make a difference in browsers that don't handle XHTML anyway. Another possibility is to use content negotiation so that URLs without a file extension, e.g.: http://dbaron.org/css2.1/tests/t010403-shand-border-00-c are sent as text/html to browsers that don't handle XHTML. 2. possibly add more <link rel="help"> tags for some of the tests. e.g. your border shorthand test should also link to the sections on the 'border' and 'border-color' property in spec, because that's really what it's testing, in fact, I might put the link to the border-color property first since that's really what is being thoroughly tested. I'm busy with prepping for my presentations at SXSW so I didn't do a detailed review and will hopefully have more comments later. Definitely a very nice start though. Makes me want to write some tests. Thanks, Tantek On 3/13/04 3:21 AM, "Ian Hickson" <ian@hixie.ch> wrote: > > On Fri, 12 Mar 2004, L. David Baron wrote: >> >> I just wrote a few tests [1] for 2.1 [2], and I'm interested in knowing >> how I'm doing, especially with regard to the guidelines [3], and >> especially from the authors of said guidelines. > > I think those are the best tests I've ever seen. > > My only comments -- and don't get me wrong; these are nit-picking > comments, the equivalent of someone saying you should fix the spelling in > a comment during a code review -- would be: > > http://dbaron.org/css2.1/tests/t010403-shand-font-00-b.xht > http://dbaron.org/css2.1/tests/t010403-shand-font-01-b.xht > Instead of "NOT be in a bold font" I would just say "be in a thin > font" or "be in a regular font weight", as negatives tend to make > understanding the test harder. > > http://dbaron.org/css2.1/tests/t040102-keywords-01-b.xht > Instead of "This should be 10em wide.", I would just say "Test". This > is for two reasons: first, it is nigh on impossible to know how much > 10em is anyway, and second, you want to have as few pass conditions per > test as possible, so that the reader can quickly establish the result. > > http://dbaron.org/css2.1/tests/t040102-keywords-00-b.xht > http://dbaron.org/css2.1/tests/t040103-case-00-b.xht > http://dbaron.org/css2.1/tests/t040103-case-01-c.xht > http://dbaron.org/css2.1/tests/t040103-escapes-01-b.xht > http://dbaron.org/css2.1/tests/t040103-escapes-02-d.xht > I haven't yet put this in the public version of the guidelines, but > there are some assumptions you can make about the test environment: > > http://cgi.w3.org/member-bin/process.cgi?method=url&url=http%3A%2F%2Fwww.hixie > .ch%2Ftests%2Fevil%2Fcss%2Fcss21%2Fguidelines%2Fguidelines.src&output=html#req > uirements > In particular, you can assume the default text color is black, so there > is no need to explicitly set it in the tests. > > http://dbaron.org/css2.1/tests/t040103-escapes-00-b.xht > You should make a third test line which makes no use of generated > content at all, and preferably change the sentence to a dummy one > such as "This is a test paragraph". > > >> A few issues where I'm particularly unsure of myself are: >> * are my title elements good? > > They seem fine to me. To be honest the titles aren't that important. > > >> * are my filenames good, especially regarding my choice along the >> a-b-c-d-e-f spectrum (for which the tests I wrote are perhaps an >> unusual case)? > > http://dbaron.org/css2.1/tests/t040103-escapes-05-c.xht > > ...should probably be a "b" not a "c", but they seem fine to me in > general. It's a judgement call, and the judgement is almost always going > to be based on how many browsers actually fail the test -- my "Evil" tests > from back in 99 when I was making the Evil Test Suite were simply the > tests that everyone failed, as opposed to the Wet Blanket Tests, which > were the tests that only a few people failed and thus "reduced the > enjoyment of others", as it were. > > >> * is my lack of consistent wording a problem? I've written multiple >> variations of the sentences "This paragraph should have a green >> background." and "The next two paragraphs should look identical." > > That all looked fine to me.
Received on Sunday, 14 March 2004 17:26:23 UTC