April 17th, 2001

User Info Icons

I got a bit bored at work, so I bashed up 3 user icons. Of course, feel free to call me names if they're no good or not appropriate.

Male userinfo icon.

Female userinfo icon.

Transgender userinfo pic. Or male.

They're all 16x16, but they don't feel as "clean" as the current (copyrighted) userinfo pic. Well, just thought I might pass some time at work. :-)
  • Current Mood
  • lurker

Just curious....

Not sure if this is the right place to ask but I have a small question.

I know you can edit custom friends groups via the web interface and create new ones. But you cannot set your posts to be visible to that group alone via the web interface. Is there any plans on allowing this to happen? I know the windows client allows it but often I am at a place where I cannot use the windows client. I know that I am probably not the only person who could use this feature.

Hardware ordered

New hardware has been ordered....
  • 2 new 18 gig 10k RPM drives for Kenny
  • 2nd 866 Mhz processor for Cartman
  • 2 new 512 MB DIMMs for Cartman, bringing it from 1G of memory to 2G. (ordered from Crucial)
Evan --- feel like doing another 3am upgrade this weekend? :-)

An explanation on why we bought the new stuff...

The main reason was because to get both database backups and the directory back up and running safely we need to get a replicated slave database on Kenny. But Kenny only has 9 gig of disk space.... not enough for the LJ database.

So, since Kenny's already going to be down for awhile, we figured we might as well upgrade Cartman before it gets slow. (We've been adding so many webservers lately, the DB is next in line to get slow). So, we decided to buy it another processor.

But, while we have it open, why not give it more memory, too? So we were going to add another 512 MB DIMM ... but shit, why not just get two? After this upgrade Cartman is entirely maxed out. The next step will be buying the Alpha, at which time Cartman becomes the slave database, and Kenny (with its new disks) becomes the image server.

I would've got the Alpha sooner, but we really don't have space for it yet... it's like 11U and we only have, well... no space in Speakeasy's little colo room. Speakeasy is expanding and building a colo room, but I'm not sure how soon.

If LJ is ready to upgrade to the Alpha (Chef) before Speakeasy's ready, I suppose we could always just move to Internap.


I have a project for anybody with mod_rewrite experience ---

I need a set of rules that will remap this:


to this:


Currently username.livejournal.com requests aren't load balanced, since we use a separate vhost for those. That sucks. (not 1 per user.... that'd really suck. we use ServerAlias *.livejournal.com)

Anybody know how to do this?

Also, could you set an environment variable if it matches, so the LJ code can make links appropriately. (if you're looking at foo.livejournal.com, the calendar link should be to foo.livejournal.com/calendar, not www.livejournal.com/users/foo/calendar, ya know?)
  • jnala

Do we need more server documentation?

In a comment over on lj_biz, tribelessnomad wrote:
Incidentally, since you're one of the few people working closely with the source code, I hope you'll give serious thought to focusing on source-code documentation instead of end-user documentation. It isn't my place to tell
you what to do -- I just want to be sure you realize that source-code documentation is VERY much needed, and there's hardly anyone available to do it.

Do people perceive such a need? We have the protocol documentation (complete, clear), the schema browser (missing many descriptions, but usually evocative column names), and the source code (written in a high-level language, in a pretty clear style). There's also the server itself, which you can run and test against.

Basically, I found it pretty easy to jump in and start hacking once someone pointed me at the source and the schema. There was definitely a learning curve, but I'm not sure whether more docs would make a big dent in it. (Well, a complete set of schema descriptions would be nice, but Brad's working on that.) More importantly, I'm not sure whether that learning curve is actually posing a barrier to people with sufficient ability and desire to work on the code.

I dunno. It seems to me like we're better off writing new code and cleaning up existing code than writing about the code (and maintaining that writing across changes). What do other people think?