Perl Software Design Weblog

Friday·06·January·2006

A conversation with Sebastian Riedel on query strings and dispatch //at 16:41 //by abe

[23:23:21] <metaperl> can the Regex attribute match the query string part also? I suppose I could write a simple app to check… [23:23:48] <@sri> no it doesn’t [23:24:22] <rconover> mod rewrite people [23:24:26] <metaperl> hmm so only path matching can occur with attributes… you do query string dispatching within a default sub by calling $c->forward as needed [23:24:36] <metaperl> rconover: what do you mean about mod_rewrite? [23:24:45] <@sri> aye [23:24:58] <metaperl> sri: what motivated that design decision? [23:25:07] <@sri> RES [23:25:09] <@sri> REST [23:25:31] <@sri> i like to think of paths as RESTish endpoints [23:25:44] <@sri> which take arguments in different forms [23:26:45] <metaperl> oh, a lot of websites have old <acronym title="Hot Tits Transport Pr0nocol (Ulrich Schwarz)">HTTP</acronym> calling conventions. that makes it harder to rework them in Catalyst [23:26:58] <@sri> example? [23:28:23] <metaperl> well, http://perlmonks.org/?node=Recently%20Active%20Threads [23:28:54] <metaperl> instead of perlmonks.org/recentlyActiveThreads [23:29:15] <@sri> why is that hard? [23:29:24] <@sri> have a default action to redispatch [23:29:46] <@sri> thats why we added default! [23:29:48] <@sri> :) [23:29:58] <metaperl> yeah, but then you end up with a default action with a huge if-then for all the different query string possibilities [23:30:13] <@sri> why? [23:30:21] <metaperl> I think there should be a FULL_URL option [23:30:29] <@sri> cat and perl are dynamic [23:30:35] <xover> mooorrrnning all [23:30:39] <@sri> just construct a action path and try [23:30:43] <metaperl> xover: good morning [23:31:07] <@sri> metaperl: it would be very easy to ass a RawRegex action type [23:31:16] <@sri> metaperl: but i don’t see a point yet [23:31:19] <@sri> add [23:31:30] <metaperl> sri: properly default is for unrecognized requests, it is serving double-duty by grokking Query strings [23:31:42] <@sri> no [23:31:48] <@sri> default is for redispatching [23:31:56] <jdv79> i skimmed jifty and it didn’t seem too appealing. did you look into it yet sri? [23:32:08] <@sri> the foo : Path is for unrecognzed [23:32:38] <@sri> jdv79: indeed, i sent some patches already :) [23:32:40] <rconover> you know having to use bind_param for blobs in sqllite sucks [23:33:12] <metaperl> “By leaving the Path definition empty, it will match on the namespace root.” [23:33:44] <metaperl> that is not the same as unrecognized is it? www.url.com/bad/path # wont match an empty Path [23:33:44] <@sri> jdv79: i see jifty as an interesting experiment, i want to see how much convention over freedom perl people are willing to accept [23:34:28] <@sri> metaperl: it is [23:34:43] <metaperl> sri: where is that documented? [23:34:49] <@sri> metaperl: everything behind the root is an argument [23:34:56] <metaperl> oh [23:34:56] <@sri> Intro.pod? [23:35:05] <@sri> for : Path [23:35:45] <@sri> the big difference between : Path and default is that Path can be local for Controllers [23:36:09] <@sri> while default always gets the whole request path in @_ [23:36:27] <@sri> so it’s made for redispatch [23:36:39] <jdv79> sri, i thought of it in a similar light. i’m just wondering if what jifty seems to do couldn’t be another layer onto of catalyst. [23:36:55] <@sri> jdv79: very easily [23:37:13] <@sri> jdv79: obra just didn’t grok cat, so he wrote his own framework [23:37:25] <metaperl> sri: you know that? [23:37:32] <@sri> what? [23:37:44] <metaperl> you his reason for writing jifty? [23:37:56] <rconover> jifty is getting good press [23:38:13] <@sri> we talked about it… [23:38:40] <@sri> top to bottom frameworks are very in atm [23:39:15] <@sri> only thing we are missing is some features in DBIC :( [23:39:48] <metaperl> brian d foy was at our local Perl Monger’s meeting yesterday. He said that <acronym title="Common Gateway Interface; Computer Generated Imagery">CGI</acronym>::Prototype is going to match the features of Catalyst - $clients are demanding it. Randal is saying the same… although draven’s journal entry (http://use.perl.org/~draven/journal/27613) shows a Merlyn wanting to ditch CGIP :) [23:40:23] <@sri> i think he will wrap CGIP around cat [23:41:01] <bda> http://video.google.com/videoplay?docid=-1032978311268695392 # Nice. [23:41:17] <@sri> what i really want is schema support in dbic :S [23:41:37] <metaperl> you mean like Alzabo reverse-engineering? [23:41:44] <@sri> naah [23:41:46] <@sri> sane [23:42:07] <@sri> think DBIx::Migration with perl code [23:42:12] <metaperl> Alzabo is a well-done product. Query formulation is irksome but the reverse-engineering and schema design tool are nicely robust