Sat Jul 07 2001 08:33:
I won a game of Illuminati with lousy cards, and lost one with really
good cards. C'est la Via Appia.
Sat Jul 07 2001 08:33:
I won a game of Illuminati with lousy cards, and lost one with really
good cards. C'est la Via Appia.
That said, I have now a set of abstractions which should work, once
I finish writing them.
Sat Jul 07 2001 08:54:
I have decided that, however much the problem of printing date
navigation bars for various levels of date navigation (eg. by year,
month, day, etc) may look like a problem easily solvable by
one uber-abstraction, it's not.
The problem is that "by year" means something different
when you're viewing a whole year of entries at once (eg. "show me the
next year") than it does when you're viewing a month's worth of entries
at once (eg. "show me this month in the next year"). The fact that
I was also trying to have the same abstraction handle both "next" and
"previous" didn't help.
Sat Jul 07 2001 10:22:
I miss strong typing.
Sat Jul 07 2001 13:32:
This entry is stored in the file nycb/2001/7/7-3 (because it's my
fourth entry today). The problem with this scheme is that sorting
the filenames, which Python can do relatively quickly, doesn't sort
the entries. I've changed it in new NewsBruiser to be nycb/2001/07/07-3,
but that still won't sort properly if there are more than ten entries
in any given day (which has happened for NYCB eight times in the past
three years). I have a custom sort which will sort that properly, but
even optimized it's 20 times slower than a straight filename sort. I'm going to leave it
at that for now, but would it make more sense to violate the zero-one-infinity
rule and make it nycb/2001/07/07-03? Then I could use the filename sort, but if for some reason
a notbook got more than 100 entries in a day, it would break.
Sat Jul 07 2001 18:24:
Da-da-da-da-da-dada! Newsbruiser Search!
The code is ugly, but I'll revisit it after I do the editor.
Sat Jul 07 2001 22:47:
It's all over but the shouting. The new NewsBruiser can do almost everything the old version can, and more (if that makes sense). It's 1000 more lines of code, but it feels like less. Tomorrow I'll be putting the finishing touches on it, and hopefully writing the docs as well.
Going to sleep now before I start making more serious mistakes.
Sat Jul 07 2001 23:14:
Oops, it's only 400 more lines of code. My mistake.
![]() | Unless otherwise noted, all content licensed by Leonard Richardson under a Creative Commons License. |