August 3rd, 2002

tom baker

e-mail address masking

First time posting here. Not sure if this is the appropriate place for a feature request.

I was looking at the way the email addresses get "scrambled" on the user info page, and I was thinking to myself, "if I was a spammer, how would I go about defeating something like this". The first answer that occured to me is "drop all whitespace". Then I looked at the HTML and found that this was wrong. Okay, so drop all tags surrounding an '@' sign? Okay, so maybe it isn't that simple.

But why doesn't LJ use a scheme like the one used on SourceForge, where they take an address such as "username@domain.tld" and literally expand it to "username at domain dot tld"? Has this ever been considered?
  • Current Music
    "Exsultate Justi", by John Williams (?) (from the "Empire of the Sun" soundtrack)

Custom style returning incorrect events

I've built a custom style (54622) that returns the latest friend post in XML syntax, for easy parseability. This is as a workaround for the limitation that the LJ protocol doesn't provide a way to retrieve other people's journal entries.

Unfortunately, this style definition seems to be exposing some bug in the LJ server software. A large fraction of the time, the post returned is not the latest one -- I regularly see it return posts that are over a week old. I've been trying this out for over a month now, and I'd say at least 1/3 of the time it returns an incorrect post.

As an experiment, I just modified the style to return the latest two friend posts. Now I seem to get the genuinely latest post first, followed by a bogus post. You can see an example here (view source to see the XML; and of course you can replace 'snej' in the URL with your own username to have it show your own friends' posts.) At the moment, this returns my latest friend post from about a half hour ago, followed by one from July 22nd. (Before I increased the number of posts from 1 to 2, the 7/22 post was the only one I got.)

Of course, this problem is intermittent, so your results may vary. But someone knowledgeable about the server software [i.e. not me] might want to look into this bug...