(I can't think of any good reason to cpntinue this discussion on
xemacs-review, so I've tried to include sufficient context and have
moved over to xemacs-beta.)
Adrian.Aichner(a)t-online.de (Adrian Aichner) writes:
>>>>> "Vin" == Vin Shelton
<acs(a)xemacs.org> writes:
Vin> Adrian.Aichner(a)t-online.de (Adrian Aichner) writes:
>> >>>>> "Vin" == Vin Shelton <acs(a)xemacs.org>
writes:
>>
Vin> Miscellaneous gripe:
Vin> - The restricted search feature of the mailing lists is quite
Vin> finicky. When I search for 'syntax table' restricting matching
Vin> files to July 2001, the search works. But if I restrict matching
Vin> files to July - Nov 2001, I get 0 matches. This wonkiness makes
Vin> the restricted searching nearly useless.
>>
>> Hi Vin,
>>
>> here's what I found:
>>
>> Querying up to 4 months works. This may be some sort of buffer
>> overflow problem.
>>
>> All I can suggest for now is to restrict searching to 4 months at a
>> time.
>>
>> Alternatively (my preference) you can do an unrestricted search after
>> you up "Max. files returned" to 1000 (even set "Max. matches
per file"
>> to 0 to get subject lines only). Does that help?
>>
>> I'll dig a bit deeper though.
>>
Hi Vin!
Vin> Thanks for checking this out. A warning on the webpage about
Vin> up to four months working is a good start, but I think we
Vin> really need to figure out how to make this rock solid. I
Yes.
Vin> also used the 1000 matches trick, but for some things that's
Vin> not enough, especially because the matches seem to get listed
Vin> from the oldest to the newest.
Vin> What is the search engine we're using?
It's wilma.
I'm currently working on texi2html to get XEmacs HTML manuals online
under
http://www.xemacs.org
I might look into the wilma problem some more.
Do you have an example (URL) where the 1000 matches don't suffice?
When I originally started to look for postings on xemacs-beta about
Matt Tucker's syntax table changes, I used the query of 'syntax table'
for every month of 2001. When that returned 0 matches, I expanded the
query to 'syntax', searching all the xemacs-beta archives. IIRC, at
that time the result did not return the 'syntax table' matches I was
expecting from the year 2001, because the number of matches ran out.
I just tried to repeat that experiment, and my results were quite
different, however:
Summary for query "syntax":Found 4005 matches in 855 files.
This time I restricted the number of matches per file to 5 - it's most
likely that on my first attempt I did not limit the number of matches
per file. So I guess the answer is no, I do not have a verifiable
test case.
Still, in my own defense, I must say that the kind of problem I
reported, where narrowing the search to a specified timeframe causes
no matches to be detected is sufficiently exasperating to the average
user that he/she will just give up. (I'm also not convinced that all
searches will return reproducible results every time, but I do not
currently have any data to back up this rather slanderous assertion.
I may attempt to come up with a test script to verify the robustness
of our search engine.)
- vin