Change the default CSS fonts for Simplified Chinese, Korean and Thai on Linux...

Change the default CSS fonts for Simplified Chinese, Korean and Thai on Linux to poplular Linux fonts. 

This CL does not change the fonts for traditional Chinese because they're the same as for English Chrome (per popular request in Mozilla bugzilla years past) and
it's not clear what to do about mscorettf (Arial, Times New Roman) we specify for most language versions of Chrome. 

BUG=20171
TEST=NONE



Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=32671

[jungshik at Google, 2009-11-17 00:03:52]

I'm leaving zh-TW alone because Traditional Chinese users expressed their wishes
to use 'Western' fonts by default (in Firefox bugzilla and that's what Chrome
Windows and Firefox Win do). 

I'm also leaving alone other languages for which MSCoreTTFs are specified. We
can replace them with DejaVu, but I'm not sure if we want that or not. 

BTW, Firefox (out of the box) has 'sans', 'serif', 'mono' for all
languages/scripts expecting them to be 'resolved' by fontconfig. 

Another BTW, do I need separate entries for 'linux2' or would 'linux' work for
it as well?

[tony, 2009-11-17 00:08:23]

http://codereview.chromium.org/259060/diff/4001/2013
File chrome/app/resources/locale_settings_ja.xtb (right):

http://codereview.chromium.org/259060/diff/4001/2013#newcode24
Line 24: <if expr="os == 'linux'">
I think this needs to be 'linux2' because that is what python uses for
sys.platform.

http://codereview.chromium.org/259060/diff/4001/2011
File chrome/app/resources/locale_settings_ko.xtb (right):

http://codereview.chromium.org/259060/diff/4001/2011#newcode4
Line 4: <if expr="os == 'linux'">
Nit: remove this empty block?

[jungshik at Google, 2009-11-17 22:54:11]

Updated the patch addressing Tony's concerns + one more (there was an
unnecessary comment in the ja file).

Evan, do you have any comment?

[Evan Martin, 2009-11-17 22:58:34]

http://codereview.chromium.org/259060/diff/10002/8008
File chrome/app/resources/locale_settings_ko.xtb (right):

http://codereview.chromium.org/259060/diff/10002/8008#newcode25
Line 25: <translation id="IDS_WEB_FONT_FAMILY">UnDotum,Baekmuk
Gulim,sans-serif</translation>
here you have commas with no spaces after but above you have a space.  all else
being equal, i think with spaces looks better.

http://codereview.chromium.org/259060/diff/10002/8009
File chrome/app/resources/locale_settings_th.xtb (right):

http://codereview.chromium.org/259060/diff/10002/8009#newcode12
Line 12: <translation id="IDS_SANS_SERIF_FONT_FAMILY">tahoma</translation>
Why lowercase these?

http://codereview.chromium.org/259060/diff/10002/8009#newcode25
Line 25: <translation id="IDS_WEB_FONT_FAMILY">norasi</translation>
Does this mean that if they don't have "norasi" we'll not have any other
fallback?

[jungshik at Google, 2009-11-18 18:03:04]

On 2009/11/17 22:58:34, Evan Martin wrote:
Can you take another look? 

Two 'formatting issues' are addressed. 
 
> http://codereview.chromium.org/259060/diff/10002/8009#newcode25
> Line 25: <translation id="IDS_WEB_FONT_FAMILY">norasi</translation>
> Does this mean that if they don't have "norasi" we'll not have any other
> fallback?

It's up to our renderer to do the font fallback, but I think it's better to list
more fonts. So, I added more Thai fonts found on typical Linux machines and
ended the list with 'sans-serif' (as done on Mac). 

I have debated whether to prepend the list with 'arial' (or 'Dejavu Sans') in
all 4 locales I'm touching because the ASCII portion of CJK and Thai fonts are
not so good looking (on the other hand, there is an issue of whether arial goes
well with CJK/Thai portion). On Windows, we didn't do, yet (although on Mac, we
do). So, I just leave them alone for now.

[Evan Martin, 2009-11-18 18:39:31]

On 2009/11/18 18:03:04, Jungshik Shin wrote:
> > Does this mean that if they don't have "norasi" we'll not have any other
> > fallback?
> 
> It's up to our renderer to do the font fallback, but I think it's better to
list
> more fonts. So, I added more Thai fonts found on typical Linux machines and
> ended the list with 'sans-serif' (as done on Mac). 

On more thought, I believe we know to use sans-serif as the worst case fallback,
so I think this maybe actually isn't needed on Mac or Linux.

I'm sorry for the confusion -- I know that contradicts what my question before
was implying...

[Evan Martin, 2009-11-18 18:40:11]

This LGTM without the sans-serif bits

[jungshik at Google, 2009-11-18 22:53:03]

On 2009/11/18 18:39:31, Evan Martin wrote:
> On 2009/11/18 18:03:04, Jungshik Shin wrote:
> > > Does this mean that if they don't have "norasi" we'll not have any other
> > > fallback?
> > 
> > It's up to our renderer to do the font fallback, but I think it's better to
> list
> > more fonts. So, I added more Thai fonts found on typical Linux machines and
> > ended the list with 'sans-serif' (as done on Mac). 
> 
> On more thought, I believe we know to use sans-serif as the worst case
fallback,
> so I think this maybe actually isn't needed on Mac or Linux.

Actually  for CJK, we fall back to sans-serif as the last resort. For non-CJK,
we fall back to serif.  That's to behave like Firefox (which has a per
script/langGroup pref. to determine whether to use serif or sans-serif in the
absence of the CSS generic family in the page stylesheet). For CJK, firefox's
default is to use sans-serif while for others, it's serif. 

See chrome/browser/tab_contents/render_view_host_delegate_helper.cc (line 126)
and the values of IDS_STANDARD_FONT_IS_SERIF in various locale_settings_*.grd
(they're true except for ko,ja,zh*). 


So, specifying sans-serif at the end does make differences for Thai among four
here because we want to use sans-serif in our domUI. 

> 
> I'm sorry for the confusion -- I know that contradicts what my question before
> was implying...

[jungshik at Google, 2009-11-20 20:19:55]

Evan, can I go ahead? It doesn't matter very much either way (sans-serif at the
end or not).

[Evan Martin, 2009-11-20 21:14:52]

On 2009/11/20 20:19:55, Jungshik Shin wrote:
> Evan, can I go ahead? It doesn't matter very much either way (sans-serif at
the
> end or not).

LGTM either way :)