Memesters - Grin with cat attached
Previous Entry Next Entry
Memesters Sep. 6th, 2003 06:19 pm
Is there a community for people working on off-site tools that interact with LJ? I don't mean via the RPC interface, lj_dev serves that purpose, I mean screen-scrapers such as marnanel's Kelvin and Joule, skx's blogmatch, ciphergoth's trustflow, my own LJtoys, and the LJ sitcom meme.

If not, is there a niche for one, to discuss tips & tricks, suggest future remote interfaces for LJ, (eg authentication, non-scraping interests gathering etc) and as a place for The Powers That Be to make comments on interfacing?

(no subject) - (Anonymous)
From: wechsler
Date: September 6th, 2003 - 10:36 am (Link)
Would you agree with my understanding that dev is for lj-internal stuff, and clients for RPC issues?

TPTB may be reluctant to create an official one, but I see no harm in creating our own.
(no subject) - (Anonymous)
From: jadine
Date: September 6th, 2003 - 03:38 pm (Link)
Is there any kind of FAQ for new people who want to try their hand at this?
(no subject) - (Anonymous)
From: wechsler
Date: September 7th, 2003 - 01:53 am (Link)
documented nicely with bloody great gaps, but still...
From: wechsler
Date: September 7th, 2003 - 03:40 am (Link)
The recent discussions on authentication would be one example, but it'd also be useful to see what other people are working on, and if they'd any ideas or insights for those or other tools.

There are also issues of handling the LJ servers appropriately - I believe that TPTB aren't overly keen on people sharing scraper code with just anyone, as this could lead to the code being used without understanding of how to do so responsibly.

Inevitably I've asked most of the questions I need to for now, but I imagine there will be others coming along later to ask them again, and I'm sure I'll find more questions in the future.
From: ciphergoth
Date: September 7th, 2003 - 03:20 am (Link)
I think this is a damn good idea.

I got a reply to my email to LJ today, from Brad, who says that userinfo.bml is expensive for LJ to generate but he's interested in adding a "web services" API specifically for applications like TrustFlow; this would mean that it could get everything it needed from the memcache without needing to hit the database. I've pointed out this post to him.
From: ciphergoth
Date: September 8th, 2003 - 02:49 am (Link)
Brad says "just use lj_dev for now. I'll post there this evening and ask about what the API should be.