222580 - smart quote entities dont' work on french user's machine

Status: RESOLVED WONTFIX

(www.mozilla.org :: General, defect, P3)

Description

[Dawn Endico]

User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.5b) Gecko/20031010
X-Accept-Language: fr-fr, fr

someone on linux who is apparently french speaking reports problems
displaying smart quotes. 

What’s New

Beyond Bliss”

my linux installation has no problem with these entities

Comment 1

[Dawn Endico]

Attachment: example of bad quote rendering

Comment 2

[Daniel Wang]

Endico, can we have a WONTFIX? Most doc people (e.g. Ian, I, Brant, fantasai,
etc.) have been using smart quotes for a while now, and lots of pages will be
affected by this. Not that I don't want to change, but we don't have many doc
maintainers.

Can someone tell me if this is a font problem, browser problem, or an encoding
problem?

Comment 3

[Dawn Endico]

i forwarded the mail to tristan. hopefully he can talk to the guy in french
and figure out what the problem is. I agree that this is probably a wontfix 
but i'd still like to investigate first. If he's running mozilla on linux
then i doubt his installation can be *that* old. Maybe there are a bunch
more like  him.

Even if we did wind up removing these quotes from the home page that
doesn't mean we'd have to remove them elsewhere, or remove the others
immediately. Developer pages are looked at by developers who are more
likely to have more modern font setups. Pages aimed at users should
work on as many systems as possible even if those systems are antiquated
and crappy. 

Also, i'm curious, what is the point of using smart quotes? why not use
regular old fasioned quotes? Is it so the beginning and end quote look
diffferent or is there more to it than that.

Comment 4

[Dawn Endico]

I said earlier that the smart quotes work fine on my system. I'd like to
note that I have truetype working on my system and have some microsoft
fonts installed. Maybe smartquotes don't work on vanilla linux boxen?

Comment 5

[fantasai]

Just for the record, I have no opinion on which set of quotation marks we're
using, and don't recall using them on mozilla.org pages in any case.
(I have switched over -- to — though.)

Comment 6

[Dawn Endico]

found a second complaint in webmaster mail from a linux user complaining
about smart quotes on home page

Comment 7

[Dave Shea]

The way I understand it, XML doesn't have many built-in character entities. Using numeric hash 
(’) instead of the named entities (") is the best way to keep your content forward-
compatible so you don't have to go through the process of converting all these again in two or 
three years.

This one sounds like a localization issue, and may have something to do with installed fonts; but 
up until now I assumed the numeric values were cool in everything reasonably modern.

Comment 8

[Dave Miller [:justdave]]

Would it be a problem to use utf8 for everything and put the actual characters
in instead of an entity?  Every browser I've seen in recent history supports
utf8 now...

Comment 9

[Mathieu Pillard [:mat]]

fwiw, I can't reproduce this using a gtk2 mozilla 1.4 build. I've tried various
combinaisons of fonts and encodings, it works fine.

(I speak french too, so I could answer to the original guy, but I'm sure tristan
will forward this to me anyway if he can't find what the problem is :-)

Comment 10

[Daniel Wang]

> Would it be a problem to use utf8 for everything and put the actual
> characters in instead of an entity?  Every browser I've seen in
> recent history supports utf8 now...

On Windows 98 Chinese, UTF8 would default text to ugly serif font. I think
Windows XP doesn't have MS Arial Unicode, so it might have the same problem.

Comment 11

[fantasai]

" is exactly equivalent to ASCII ".
The named entity is primarily for use in attribute values, which can't take the
ASCII quotation mark directly. (It's used as the value delimiter.)

The question is whether or not to use UNICODE curly quotes. (The character
entities are essentially referring to the UNICODE codepoint.)

Comment 12

[chris hofmann]

p3

Comment 13

[Tristan Nitot]

Dawn, please send me ([e-mail]) the contact details so I can contact
the reporter in french.

Comment 14

[Dawn Endico]

I already forwarded it to you there. maybe it got filtered to junk?
Just forwarded again.

Comment 15

[Jungshik Shin]

> On Windows 98 Chinese, UTF8 would default text to ugly serif font. I think
> Windows XP doesn't have MS Arial Unicode, so it might have the same problem.

Neither of this is an issue at all if you specify lang or xml:lang. 

That is, with <body lang="en" xml:lang="en"> or equivalent in the page, Mozilla
picks up fonts for 'Western European' lang group whatever encoding the page is
in. The font selection based on the character encoding is the last resort (which
breaks down for UTF-8 pages) when no better information is availbale. With
lang/xml:lang, Mozilla doesn't have to take that route.

As for the problem reported, we have to know which build of Mozilla (the default
build from mozilla.org that uses X11 core fonts and Freetype2 or Xft build that
makes use of Xft) was used. Anyway, it's odd that those quotes are replaced by
characters having nothing to do with quotes. If it's a font issue, I'd expect
them to be rendered as '?' (in the default build [1]) or 'domino' (Unicode
codepoint enclosed by a square box : minifont in Xft)

[1] The default build 'transliterates' characters (for which there's no font on
the system). '?' is used when no 'approximation' is possible. I thought the
transliterator would replace smart quotes  by ASCII quotes if there's no font to
render them. The problem might be due to an error in the transliterator code.  

Comment 16

[Mathieu Pillard [:mat]]

I just posted a little something on www.linuxfr.org to see if we can get other
linux people to reproduce this bug (I still don't have it on my linux box)

(For those who speak french, its here: http://linuxfr.org/~remat/6815.html)

Comment 17

[Dave Miller [:justdave]]

reassigning endico's bugs to default owner

Comment 18

[David Boswell]

I'm closing this as wontfix based on earlier comments and based on the fact that this is related to a previous version of the site.  If people are still seeing this issue though, please reopen.