Grr

[cjsmith]

I'm getting really tired of my Friends page being seriously screwed up by badly-formatted entries people make. Today I'm guessing it's quiz results. An entry I can't see makes everything below itself almost completely unreadable.

Friends filters work, but to use them I have to find out who the problem child is. This is particularly difficult to narrow down when multiple people have the same problem; searching the solution space becomes slow! The slowness is aggravated by LJ's "optimization" on Friends page loading, such that I have to hit "reload" to get actual results from each experiment. (I wasted a lot of time getting nonsensical and conflicting results before I learned this.)

Today two or three iterations with existing friends groups showed me a subgroup of ten people I could remove just to get a viewable page. I re-added them one by one, a process taking about fifteen minutes with all the clicking around and reloading, and came up with two people filtered out.

*sigh* And I wonder why LJ takes up so much of my time.

Comments

[hitchhiker.livejournal.com]

I was thinking more of a serverside option. Basically, the LJ pages you see are dynamically generated anyway, and bad HTML inside an LJ-cut entry doesn't affect the main page, so it would solve that particular problem neatly.

Besides, it'd be wonderful to have it fold per-user, frinstance, for all those people who ought to cut and don't, or per community, so that you could scan community posts by first line and expand only the ones you wanted to.

[joedecker.livejournal.com]

Sure--but it'll be peppier if you could do it client-side. Hmmm.
The 'deals with bad HTML' is a nice thing, fer sure.

Hmmm, there might be some tricks one could play with rewriting
the page using document.write(), and only writing the 'open' bits
and the infrastructure code but not the closed stuff. I think you
have to have JS code which writes itself to do this, which might
make it more challenging, but there might be a cute hack short
of self-expressing programming.