Less Borken Code.

[swestrup]

I've spent the entire day trying to get a few simple routines to work correctly. Now admittadly, my brain has been full of dead insects for most of the day, and its been hard to concentrate. Its also a fact that I'm dealing with 2 separate pieces of code in transition:

1.  XEmacs has got a bunch of legacy code that is kinda wonky, and so is disabled by default. One of which is that when you do a non-incremental search, it doesn't highlight the results, although it does for incremental. It turns out that this is because LOTS of code uses the incremental search internally, and none of its knows about result highlighting. So, if you remove the code that deliberately breaks result highlighting, the results get ugly. It took me all day to find out what the approved mechanism is for implementing result highlighting is, and I got the editor really sick a few times testing ideas, but the final result work fine, don't step on anyone's toes and are only a few dozen lines long. Most of those lines are repetitive as well, so I'm sure I could get it to a handful of lines, if I knew elisp better. Hmm. I suddenly wonder if there is an Emacs/XEmacs/Elisp community on LJ?
2. Apache states umpteen-gazillion times in its docs that correct configuration and operation requires a clear knowledge of the distinction between the webspace and filespace names for objects. This is all fine and dandy, but the vast majority of the modules (with the notable exception of mod_rewrite) confuse the two. In fact, EVERY module that could be written in a way that confused the two, was. What's worse, is that there are actually more naming levels than 2. In addition we have at least one other: synthetic filespace (since there are filename->filename remappings). Everyone says mod_rewrite is hard to use (and BOY IS IT EVER!), but the main reason for this is the URL/FILENAME confusion. Every command and option in mod_rewrite needs to be annotated as to when they work on URLs and when they work on FILENAMEs and at what stage of the Apache file-serving process they get invoked. Its possible to have a one-line command that deals with URLs and FILENAMEs at 3 or 4 different phases in the Apache serving process, and the only way to know is to try things and see what happens. Its not like its even anottated in the damn debug logs!

So, given all that, I've decided to be tentetively proud of the fact that I have, indeed, gotten search to work the way I want, as well as finally getting the rewrite rules to work in all reasonable circumstances. I may just call it a day now, but since I'm not feeling too well, I'm not even sure if I'll manage to sleep.