Sanskrit: Devanagari UTF-8 browser tests

[chicken]

kerri9494 said I should spam this stuff into my LJ, so here goes.

Edited at 20:33 EDT to add that of course kerri9494 has already tested all the Windows browsers (that run under XP, of course) and has those results sitting on her hard drive.

Any Linux users out there? paulv? If so, when you have a minute, please report your Linux browsers' behaviors, I'm curious.

So anyway, I was previously aware of the lameness of most browsers when it comes to rendering many of the more obscure UTF-8 characters.

One of our projects is all about the Sanskrit, which is written in a character set called Devanagari.

This is represented in Unicode in the U+0900 to U+0970 range (aka the 0xE0A480 to 0xE0A5B0 range in UTF-8 hex).

So I decided to put the "lame browsers" theory to the test.

I tested lots of Mac browsers for problems vis-a-vis Devanagari UTF-8, etc.

I took the word "evaM" (which means 'thus', I think) as the test. It ought to look like this in a browser that understands utf-8:





The dot in red there is the 'M', which gets combined over the 'va' sound. If I put a <span style="color: #FF0000;"> around it to highlight it (such as when a student left it out, so we can show them the error), some browsers freak out even more than they already do just trying to handle the utf-8.

Also, some browsers don't recognize or deal with XML files (esp. ones called something.xml), so that is also reflected in some of the results.

Clearly on a Mac, Opera rules for this sort of thing.

Test Results:

P.S. -- Some of the 'dot to side [right]' or 'dot on left' may
actually vary depending on the text zoom in various of these
browsers. No time to nail that down definitively right now.

i.e. the dot moves according to how large you make the text, in some
browsers (esp. and at least Safari).

http://carole.fates.org/Sanskrit/Test_Files/utf8-Test.xml
http://carole.fates.org/Sanskrit/Test_Files/utf8-Test.html
http://carole.fates.org/Sanskrit/Test_Files/utf8-Test-NoMargin.html

Note I made the font size for these tests 5.1em, or about five times bigger than the default font size in your browser. This is so that the "dot-to-the-side" (left or right) mistake gets minimized in most browsers, and to make the character more visible. However, if you make the font smaller in your browser (View->Text Zoom->minus or smaller type thing), you'll see how lame it gets in a lot of browsers.

F-name	Platform	Browser	Version	Behavior
utf8-Test.xml	Panther	Firefox	0.8	Dot to the side, not red
utf8-Test.xml	Panther	Camino	0.8b	Dot to the side, not red
utf8-Test.xml	Panther	Opera	7.51	Dot red, on top correctly
utf8-Test.xml	Panther	I.E.	5.2	Garbage (doesn't recognize UTF-8)
utf8-Test.xml	Panther	Safari	1.2.2	Garbage (doesn't recognize UTF-8)
utf8-Test.html	Panther	Firefox	0.8	Dot red, on top correctly
utf8-Test.html	Panther	Camino	0.8b	Dot red, on top correctly
utf8-Test.html	Panther	Opera	7.51	Dot red, 'e' and 'va' not visible
utf8-Test.html	Panther	I.E.	5.2	Garbage (doesn't recognize UTF-8)
utf8-Test.html	Panther	Safari	1.2.2	Dot red, but on LEFT (?!)
utf8-Test-NoMargin.html	Panther	Firefox	0.8	Dot to the side, but IS red
utf8-Test-NoMargin.html	Panther	Camino	0.8b	Dot to the side, but IS red
utf8-Test-NoMargin.html	Panther	Opera	7.51	Dot red, on top correctly
utf8-Test-NoMargin.html	Panther	I.E.	5.2	Garbage (doesn't recognize UTF-8)
utf8-Test-NoMargin.html	Panther	Safari	1.2.2	Dot to the side, but IS red

Comments

[mmcirvin.livejournal.com]

oh, yeah, and Safari displaced it to the right in the "no margin" test.

[jwgh]

I'm seeing the same thing with Safari. Blowing the font size up real big makes weird things happen -- eventually the red dot appears to 'wrap' to the next line, so it's actually under the rest of the glyph.

[chicken-cem.livejournal.com]

All weird and intriguing, I will diff what you have with more tests on a different Panther machine at work on Monday (got Jaguar at home, bleh).

[jwgh]

I'm using Safari 1.02 on Panther, incidentally. I wonder if what font you're using has an effect?

Oooh, just tried looking at the pages in FireFox 0.8 and, damn, some goofy stuff sure happens. I may have to take some screenshots.

[jwgh]

Well, now that I've looked at the source of the pages I guess this isn't as surprising as I thought it was, but anyway I collected some Firefox screenshots for you.

I still think this is pretty cute: