Tim Horton wrote:
Tim Vera On Jul 8, 2009, at 17:53, Magnus Therning <magnus therning org> wrote:Josselin Mouette wrote:Le jeudi 02 juillet 2009 à 18:03 +0300, Xan Lopez a écrit :Open bugs with the extensions and me or someone else will surely port those, they sound pretty simple.[..]I have to say I’m also not fond of the JavaScript idea. Is it possible now to access the DOM tree from the extensions? If it’s still not, it sounds like a big regression from the Python extensions, from which you can use the tons of existing Python modules.Yes, I second that. I'm currently struggling with rewriting Epilicious (with a renaming in the process) in Seed. The lack of basic things like a dictionary type is really quite a nuisance. The same goes for a set type. I'm sure there are quite a few other things that I'll miss as I go along.Agreed. Make a list and we'll see what we can do. A library of types goes against the whole minimal-platform thing Robb was preaching when we first sat down and wrote Seed, but we've already reached beyond that, so I don't see any reason not to work on it. I'm assuming that your complaint about heaviness of dropping to glib is just over the syntax, right, and that attractive JavaScript-looking wrappers would be enough? That's my biggest thing, I think... The syntactic heaviness, not performance/memory. I'm going to play around after ephy seed shell.
Yes, the heaviness is regarding syntax. It just feels wrong to have to specify things like comparison functions when creating a dictionary, and then using methods like add() and lookup() instead of simply indexing. I'd argue that minimality is good, but only on some levels. Pleas do keep the Seed _interpreter_ small, with a small set of _built-in_ types. But on top of that there is a real need for a rich set of modules/libraries. Without it Seed/JS simply won't take off in Gnome... well, at least I won't be very happy with having to rewrite "basic functions" over and over. I fear that would push Seed in the same direction as Guile, OCaml, etc, i.e. to obscurity. I'll put together a list as I come across stuff that I miss :-)
On the whole I'm so far not completely against Seed as a replacement of Python as an extension language. However, at the moment the missing pieces are painfully obvious. In the long term Seed does offer a chance of better support and better integration (e.g. I hope I'll be able to stop re-compiling epiphany for AMD64 just because the Python bindings break when compiled with optimisation on that platform, a bug that's been known for years now). Oh, BTW, I could _really_ need a Seed Console extension! :-)I have a working seed console (sorry about the slow, I've been in and out of home a lot) except for one important thing: redirection of stdout. So like ... Seed's print is redirected, and the response from the repl, and exceptions, but not everything else (any printing from C). When I get home (iPhone = not the best Seed hacking device!) again, I promise I'll finish implementing history and send it to you in the state it's in, as it's already plenty useful!
Cool! :-) /M -- Magnus Therning (OpenPGP: 0xAB4DFBA4) magnus＠therning．org Jabber: magnus＠therning．org http://therning.org/magnus identi.ca|twitter: magthe
Attachment:
signature.asc
Description: OpenPGP digital signature