Mercurial > ~dholland > hg > ag > index.cgi
comparison doc/admin/bugs.txt @ 0:13d2b8934445
Import AnaGram (near-)release tree into Mercurial.
author | David A. Holland |
---|---|
date | Sat, 22 Dec 2007 17:52:45 -0500 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
-1:000000000000 | 0:13d2b8934445 |
---|---|
1 Some known bugs that can be described concisely. | |
2 (See also POSSIBLE BUGS in todo-small.txt.) | |
3 | |
4 ------------------------------------------------------------ | |
5 | |
6 - It doesn't seem to issue a #line directive for the leading block of | |
7 embedded C. This might have been intentional, but seems wrong to me. | |
8 | |
9 - It should warn if an empty character set (which can never be | |
10 matched) appears in a production. | |
11 | |
12 - If you do "[ sticky { foo } ]" where foo is otherwise unreferenced, | |
13 you'll be told that foo is both unused and undefined. This is | |
14 needlessly confusing. (May be true with "lexeme" also.) | |
15 | |
16 - Building parsers from files with dashes in their names seems to | |
17 result in uncompilable output. | |
18 | |
19 - There's a "line numbers path" setting, that I didn't know about | |
20 until I just found it. But unlike "parser file name", you can't put | |
21 '#' in it to expand to the basename of the file. | |
22 | |
23 - If AG crashes badly, it calls bailOut(). In the windows version, | |
24 this pops up a dialog box, which is fine for the GUI version. But it's | |
25 also true for agcl.exe, and I doubt this is really desirable. |