From: Philippe Verdy (verdy_p@wanadoo.fr)
Date: Fri Mar 17 2006 - 10:10:48 CST
From: "Antoine Leca" <Antoine10646@leca-marti.org>
> Philippe Verdy wrote:
>> Just try to view this French Wikipedia page
>>
> http://fr.wikipedia.org/wiki/Table_des_caract%C3%A8res_Unicode_%280000-0FFF%29
>
> By the way, this page is far too large to my taste (>340 KB).
I've tried to make this page smaller, but there's a limitation in Wikimedia that does not allow inserting custom CSS stylesheet in the document. But beforethe changes, the page was still MUCH larger,due to the presence of an unnecessary link on each displayed character. And you could not access it per block without first loading a page trying to show the complete BMP in a single table.
On IE and Firefox the page loads in seconds and renders quite fast, including from browser's cache when navigating.
> Here, this page does not display correctly... because I am using a
> lightweight browser (Opera), which is not supposed to perform correctly in
> the most complex situations.
It's been long since I tried Opera and abandonned it, since it's not really more lightweight than IE (in terms of processing resources), and is really so poor with internationalization, has too many unrecognized quirks face to HTML and CSS standards, and that its support is minimalist (does it have the value we haveto pay for it?). Mac users will prefer Safari, Windows and Linux users have the choice for other browsers.
I really like Firefox for it better overall international support and better stability of its CSS rendering engine (IE7 Beta still has too many rendering bugs, including new ones like arbitrarily clipped text in centered cells or divs, and it still does not allow setting fonts properly for some scripts encoded since long in Unicode 4, or even before like Indic scripts ; anyway, IE6 loads fast on Windows and is stable, provided that the system is well protected from pollution by third-party softwares, or malwares).
Of course, the best platform for sending accurate international text is still PDF documents, but Adobe Reader has its own stability problems in its plugin for browsers, which hangs so often for unknown reason.
> When I switch to another browser, namely plain old IE6sp1 (on W2000sp4 with
> a somewhat updated Uniscribe and the relevant fonts), it displays perfectly
> OK...
> Similarly, almost perfectly (only lacking the codepoints added in TUS4, as
> expected) displayed on a freshly installed 2003 server sp1.
>
> In fact, I do not have one to test here easily, but I expect a regular
> (unpatched) client using Windows XP to performs OK as well (perhaps
> providing the Gurmukhi etc. fonts have been installed if it is not the
> default).
>
>
> <OT>
>> Look at the places where a bold "Note :"
>> indicates various rendering bugs found most often in Internet
>> Explorer or in Firefox.
>
> Such nominal statements would not be proper of the NPOV of the wikipedia, in
> my humble opinion.
> </OT>
There's no NPOV here. It just says "some browsers" without naming them. This statement is there to warn readers that some parts of the tables may not display correctly. Note that somedefects are intermittent with the same browser on the same installed host.
Notably in Firefox, scrolling down and up the page,withouteven reloading it, can sometimes change the final rendering, and correct some noted defects (for example the Coptic block appears with wrong encoding order of letters the first time it appears on the screen, but scrolling up and down changes the font style, and the letter ordering. It seems that Firefox provides alternate code mappings for emulating characters that it can't temporarily find in available fonts, but later when fonts are located the internal code remappings are no more used, or the ordering of fonts in preferences is not stable for the same page. This is apparentky a bug reported to the Mozilla team)
This archive was generated by hypermail 2.1.5 : Fri Mar 17 2006 - 10:16:29 CST