W3C home > Mailing lists > Public > public-css-testsuite@w3.org > March 2013

Re: Text-orientation and text-underline-position tests (text-orientation-mixed-001)

From: Kazuaki Takemura <takemura@networksoft.co.jp>
Date: Mon, 18 Mar 2013 13:24:33 +0900
Message-ID: <51469701.4000709@networksoft.co.jp>
To: Gérard Talbot <css21testsuite@gtalbot.org>
CC: Public CSS test suite mailing list <public-css-testsuite@w3.org>
Hi Gérard,

Thank you for the review.
(2013/03/15 16:29), "Gérard Talbot" wrote:
> Kazuaki,
>
> http://test.csswg.org/source/contributors/ktakemura/submitted/text-orientation-mixed-001.xht
>
> <p>Test passes if both "Text sample" rectangles are
> <strong>identical</strong>: Only "T" should be displayed vertically, and
> "ext Sample" should be displayed sideways.</p>
>
> Suggestion (more precise):
>
> <p>Test passes if both "Text sample" rectangles are
> <strong>identical</strong>: only "T" should be displayed vertically and
> "ext sample" should be displayed sideways.</p>
Okay, It was fixed.

>
>
> line 27: margin: 10px;
>
> You add 10px on all 4 sides... but the reftest only adds margin-right of
> 10px.
Okay. I corrected to "margine" from "margine-right".

>
>
> This text-orientation-mixed-001.png image has smooth edges, anti-aliased
> edges but the div#vertical has aliased edges. I have right now no idea
> how to work around this, except by using Ahem font. Some browsers like
> Opera under Linux (and only under Linux) will anti-alias text by
> default.
Processing of anti-aliasing is system dependence.
Therefore, I think it is better to turn off anti-aliasing. Then, I 
checked WorkingDraft of Font-Module.
At old WorkingDraft, although it seems that there was a property of 
font-smooth, there is no now.


> Over here, between DejaVuSerif-webfont.woff and the
> text-orientation-mixed-001.png, the "e" glyphs are different, slightly
> different but noticeable; the "a" are also a bit different.
> This could be because DejaVuSerif-webfont.woff is based on version 2.29
> (2010-12-23) while, on my system, when creating that image with, say,
> GIMP 2.8.2, I would be using DejaVu Serif version 2.33 (2011-02-27)
> which could be different. Also, the tool by which we create .woff could
> be inaccurate too.
It checked also here. It seems that it is different also by webkit and 
opera.
Although a system cause can be guessed, I think that it is very 
difficult to create exact reftest
  in the present condition.

> "
> I would suggest being cautious about using subsetting tools if you don't
> know much about their quality or where their deficiencies lie. There are
> many tools that will create all sorts of funky problems (bad cmaps,
> incorrect or inconsistent tables, etc.).
> "
> http://lists.w3.org/Archives/Public/public-css-testsuite/2013Feb/0037.html
>
> and this DejaVuSerif-webfont.woff is a subset of DejaVu Serif (only
> English glyphs).
>
>
> Kazuaki, (it's 03:30 am here) I will check your other tests tomorrow. I
> won't go to the meeting I was expecting to; so I should have time for
> your tests tomorrow.
>
> Gérard

I sent e-mail to Mr. Ishii, but there is still no replay.

Regards.
Takemura
Received on Monday, 18 March 2013 04:26:31 GMT

This archive was generated by hypermail 2.3.1 : Monday, 18 March 2013 04:26:33 GMT