#16 Multiple bugs processing ht://Dig

[Thomas Bryan]

Submitted by Geoff Hutchison on Tue, 13 Jul 1999
22:24:20 -0400 (EDT)

Hi,

First off, let me say I'm sure you're awefully
busy with taking over as maintainer--I know the
feeling. I took over the ht://Dig search engine
last summer and the first bit can be tough.

I'm interested in using PERCEPS to maintain
documentation for the ht://Dig project's C++
classes. Right now it looks like 90% of what we
want is there and it's already very
useful. However, certain files seem to hang the
parser, it doesn't warn you when the output
directory doesn't exist (oops!) and there don't
seem to be ways of outputting the source filename
(since we have several source directories) or the
comments at the beginning of the source file
(where we usually put the class description).

I can send you the source that hangs the
parser. Are we missing the outputs for the source
file comments and the filename?

Anyway, for the difficulty of the problem and the
quality of the present solution, congratulations
are in order!

-Geoff Hutchison

[Thomas Bryan]

[Thomas Bryan]

Logged In: YES
user_id=1197

Submitted by Geoff Hutchison on Wed, 14 Jul 1999 13:20:11
-0400

Tom Bryan wrote:
>> However, certain files seem to hang the parser,
> Not surprising. :-\

No, it's not. Obviously, in the meantime I'm
parsing everything else. A tar.gz of those files
is attached.

> Thus, if you start your file with a short
> description (a '//:' comment), all of the
> comments at the top of the file should become
> "attached" to the first C++ element that PERCEPS
> find in the file. If that's the

Would this include #include and #ifdef and the
like? e.g.

//: Interface to the word database. Currently this
writes...
//
#ifndef _WordList_h_
#define _WordList_h_
... (more defines and a few blank lines)
class WordList
{

> arbitrary class (or other C++ element) in his
> code. Of course, then there would be problems
> with how to handle multiple comments attached to
> one class.

I'd probably just catenate them. Not the best
solution, but...

> The current release is 100% Mark Peskin. I
> haven't made any releases since I took over a
> few weeks ago. Actually, I'm having trouble
> deciding how often to post revisions. I've been
> meaning to code a few bug releases while I work
> on a rewrite, but I haven't had time in the last
> few weeks.

I generally limit bug-fix releases to 1 or 2,
unless it's absolutely critical. Otherwise it
starts taking away from the development. It's
unbelievable how much time it can take to make a
release properly--make sure all the version
strings are updated, update a web page or two,
announce it to various lists and pages...