HP Forums

Full Version: Mathematical constants and computations, by Xavier Gourdon and Pascal Sebah
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
http://numbers.computation.free.fr/Const...tants.html

This is another quite interesting site, with quite interesting articles which seems to be abandoned for quite a long time...

I have sent to authors a message few years ago, reporting that almost all formulas on site are broken, however no respond. It seems that site is last time updated in 2010.

I believe this also deserve to be preserved and recreated, if it is indeed abandoned by authors, before disappear...
(10-05-2018 06:10 PM)sasa Wrote: [ -> ]http://numbers.computation.free.fr/Const...tants.html

This is another quite interesting site, with quite interesting articles which seems to be abandoned for quite a long time...

I have sent to authors a message few years ago, reporting that almost all formulas on site are broken, however no respond. It seems that site is last time updated in 2010.

I believe this also deserve to be preserved and recreated, if it is indeed abandoned by authors, before disappear...

The formulae look broken due to use of the old Symbol font, which has its own, peculiar character encoding. The whole site would probably need an overhaul to be more compatible with modern browsers.

If it disappears, at least it still seems to be available via the Internet Archive.
(10-05-2018 06:57 PM)ijabbott Wrote: [ -> ]The formulae look broken due to use of the old Symbol font, which has its own, peculiar character encoding. The whole site would probably need an overhaul to be more compatible with modern browsers.

Taking a look at http://numbers.computation.free.fr/Constants/Pi/pi.html I do indeed see p instead of π, and in the HTML, there is <FONT face=symbol>p</FONT>. I wonder why that doesn't render as intended, though, because the Symbol font is present on my computer. When I select it in MS Word, and type a lowercase p, I see a lowercase π, as expected. Why does that work in Word but not in the browser?

Decoding those formulae would be a lot of fun /s note how the divisions are constructed using table objects and <hr> tags...

EDIT: When I copy the text from the browser and paste it into Word, it renders as intended.
(10-05-2018 07:58 PM)Thomas Okken Wrote: [ -> ]
(10-05-2018 06:57 PM)ijabbott Wrote: [ -> ]The formulae look broken due to use of the old Symbol font, which has its own, peculiar character encoding. The whole site would probably need an overhaul to be more compatible with modern browsers.

Taking a look at http://numbers.computation.free.fr/Constants/Pi/pi.html I do indeed see p instead of π, and in the HTML, there is <FONT face=symbol>p</FONT>. I wonder why that doesn't render as intended, though, because the Symbol font is present on my computer. When I select it in MS Word, and type a lowercase p, I see a lowercase π, as expected. Why does that work in Word but not in the browser?

Decoding those formulae would be a lot of fun /s note how the divisions are constructed using table objects and <hr> tags...

EDIT: When I copy the text from the browser and paste it into Word, it renders as intended.

I do see the "pi" symbol in my browser (latest chrome on win10).
(10-06-2018 06:00 AM)DA74254 Wrote: [ -> ]I do see the "pi" symbol in my browser (latest chrome on win10).

This perhaps is only fortunate with the lastest Chrome on Win10...

However, I use Linux only (Ubuntu) and every browser I tried failed, including latest Chrome. Exactly as Tomas wrote: copying formulas to LibreOffice Writer show the symbols correctly, however when that is saved back to HTML, any tested browser fails to show it correctly.
Dunno about you lot, but there used to be a truetype dir in /usr/share/fonts. I don't know what happened to it, if it doesn't still exist (it does on my Ubuntu but not on my Fedora). If that exists anywhere, shove the symbol.ttf into that, rerun whatever the right command is to resync the font dir, and hey presto, symbol. In my case, I happened to have the correct fonts in /usr/share/wine/fonts already, so I simply copied those (as root, of course) into /usr/share/fonts/truetype (after creating the directory), and fired up the browser. I went to the link in the original post, clicked on "Constants" and ... yep. They all seem to render. The root symbol looks a little weird, but I can live with that. You may even be able to get the correct fonts from the ttf-mscorefonts-installer (Ubuntu/Debian) package.

(Post 301)
(10-05-2018 07:58 PM)Thomas Okken Wrote: [ -> ]
(10-05-2018 06:57 PM)ijabbott Wrote: [ -> ]The formulae look broken due to use of the old Symbol font, which has its own, peculiar character encoding. The whole site would probably need an overhaul to be more compatible with modern browsers.

Taking a look at http://numbers.computation.free.fr/Constants/Pi/pi.html I do indeed see p instead of π, and in the HTML, there is <FONT face=symbol>p</FONT>. I wonder why that doesn't render as intended, though, because the Symbol font is present on my computer. When I select it in MS Word, and type a lowercase p, I see a lowercase π, as expected. Why does that work in Word but not in the browser?

On my Linux machine, "symbol" gets matched to the "OpenSymbol" font. It has the Greek characters and various symbols, but at the standard, Unicode code points. It doesn't have any glyphs at the ASCII (roman) letter code points (where the Adobe "Symbol" font puts the Greek characters), so p gets displayed as p by the Chrome and Firefox browsers.

My Windows 7 machine has the "Symbol" font from Monotype (I think if is preinstalled) and has the Greek characters at the ASCII letter code points (I guess it's a clone of Adobe's "Symbol" font), so the p is displayed as π by the Chrome and IE browsers, though strangely, Firefox persists in showing it as p.

I guess the web-pages could be converted to UTF-8 by transliterating the parts spanned by the <font face="symbol"> tags. But the formulae constructed using tables would need to be manually converted to use MathJax or MathML.

(EDIT: The encoding is shown on Wikipedia: Symbol_(typeface)#Encoding.)

Quote:Decoding those formulae would be a lot of fun /s note how the divisions are constructed using table objects and <hr> tags...

I noticed that. I guess more modern math rendering wasn't an option for them at the time, especially considering the pages appear to be constructed with Microsoft FrontPage 4.0 and 5.0.

Quote:EDIT: When I copy the text from the browser and paste it into Word, it renders as intended.

I'd hope so, since it was generated with Microsoft tools in the first place!
(10-06-2018 07:35 AM)brickviking Wrote: [ -> ]Dunno about you lot, but there used to be a truetype dir in /usr/share/fonts. I don't know what happened to it, if it doesn't still exist (it does on my Ubuntu but not on my Fedora). If that exists anywhere, shove the symbol.ttf into that, rerun whatever the right command is to resync the font dir, and hey presto, symbol. In my case, I happened to have the correct fonts in /usr/share/wine/fonts already, so I simply copied those (as root, of course) into /usr/share/fonts/truetype (after creating the directory), and fired up the browser. I went to the link in the original post, clicked on "Constants" and ... yep. They all seem to render. The root symbol looks a little weird, but I can live with that. You may even be able to get the correct fonts from the ttf-mscorefonts-installer (Ubuntu/Debian) package.

(Post 301)

That's an idea. The command to update the Fontconfig cache (used on Freedesktop based systems) is fc-cache, probably with the -f option.

The symbols look a bit different in the Wine version of symbol.ttf as it's not the same as the Windows version of the font. It's the "Wine Symbol" font rather than the Monotype "Symbol" font used by Windows.

FWIW, the Microsoft TrueType core fonts don't include "Symbol".
Reference URL's