On Mon, Nov 1, 2010 at 14:47, Don Stewart wrote:

magnus:

...

I know there's a .cabal file for the latest version of HP somewhere, but I can't coerce Google into finding me a link that actually works. Furthermore, the following page:

http://hackage.haskell.org/platform/contents.html

does list all the contents, but to my big surprise it doesn't link to the specific versions of the packages for HP, instead it links to the latest version found on Hackage.

Would someone with the power to make changes on the HP pages *please* make it as easy as possible to find the *exact* specification of what HP contains?  Please, pretty please with sugar on top.

No, a changelog entry (http://hackage.haskell.org/platform/changelog.html) is not very helpful (why the HP front page links to it I can't understand).

Going via the Haskell wiki (http://www.haskell.org/haskellwiki/Haskell_Platform#What.27s_in_the_platform) to find a link to the .cabal (http://code.haskell.org/haskell-platform/haskell-platform.cabal) is not that user friendly.  It's even worse that the latter link doesn't seem to work at all at the moment.

(The short irritated tone in this email accurately shows my desperation with the situation: I thought I would be able to find this information with only 5 minutes to spare before my next meeting.)

Currently, the versions are specified in the .cabal file. A script is used to generate the changelog page (diffcabal, iirc).

I'll generate a spec page from the .cabal file this week sometime.

Ah, excellent. Sorry for the rather rant-y email before. I've now been to the meeting and managed to calm down a bit :-)

P.S. better sent to the haskell-platform@ list

Yes, of course. I'm now trying to fix that by cross-posting, let's hope I won't get too many angry emails about that ;-) /M -- Magnus Therning                        (OpenPGP: 0xAB4DFBA4) magnus＠therning．org          Jabber: magnus＠therning．org http://therning.org/magnus         identi.ca|twitter: magthe

On 05/11/2010 02:59 PM, Don Stewart wrote:

The changelog now lists all the versions:

http://hackage.haskell.org/platform/changelog.html

This is quite optimal. It would still be nice if one could easily answer the question "which HP release was the one that contained process-1.0.1.1", but this is a step in the right direction. Would it be hard to replace "->" with a real Unicode arrow character?

On 2010-11-05 21:05 +0000, Stephen Tetley wrote:

On 5 November 2010 20:08, Andrew Coppin wrote:

...

Would it be hard to replace "->" with a real Unicode arrow character?

It should be quite easy - whether a given font has an arrow readily available is a different matter. It might be be simpler to drop into the Symbol font (should be present for all broswers) and use arrowright - code 0o256.

Except that the "Symbol" font family is not available in all browsers. -- Nick Bowler, Elliptic Technologies (http://www.elliptictech.com/)

On 05/11/2010 09:05 PM, Stephen Tetley wrote:

On 5 November 2010 20:08, Andrew Coppin wrote:

...

Would it be hard to replace "->" with a real Unicode arrow character?

It should be quite easy - whether a given font has an arrow readily available is a different matter.

I can't remember the last time I saw a browser that couldn't do this. There /are/ symbols that don't work reliably, but the basic arrow symbols seem to be pretty well supported.

On Nov 6, 2010, at 3:56 AM, Stephen Tetley wrote:

Modern browsers might add in arrow from a different font if it is not present in the one chosen by the web page author - I suspect this is happening on this page where the arrow "looks wrong" typographically:

I don't think that's what's going on. Notice that the font used in that page is fixed width. The arrow looks wrong because it has to fit in a fixed width space. Fixed width arrows all look stubby, even if they are styled like the rest of the font. Very few programming fonts have arrows or other symbols. When I played around with UnicodeSyntax, I tried lots of different fonts. The only one I found that is halfway readable is GNU Unifont (at unifoundry.com)

Stephen Tetley wrote:

I thought I read that Firefox does a font swap if it can't find a glyph, but thinking about it myself I can't see that this would make sense - Firefox would have to know an awful lot about the OSes fonts to know if they have "missing" glyphs.

You're pretty much right, Firefox does read the font metrics information of _all_ fonts on the system at startup [1]. It chooses substitute glyphs based on which fonts have the required glyph and matching metrics. This font substitution can cause odd looking results if you use odd characters in a run of ascii text[2], but seems to do a reasonable job with displaying mathematical symbols. John [1] Yes, this can cause significant slowdown for users with many fonts: https://bugzilla.mozilla.org/show_bug.cgi?id=600713. I think Chrome just relies on a set fallback font with known wide unicode coverage. [2] See what can happen if you cut and paste fancy content into the web browser, but only have ligatures defined in some unusual font: http://www.undermyhat.org/blog/wp-content/uploads/2009/09/ScreenShot562.png