Elsner �����(�):
> Aspseek 1.2.5 installation and configuration is much the same as 1.2.4, OK.
>
> My test environment (P3, 256MB RAM), not working because of UtfStorage ...
> http://pandora.rz.uni-osnabrueck.de/cgi-bin/s.cgi (simple search)
> http://pandora.rz.uni-osnabrueck.de/cgi-bin/s2.cgi (advanced search)
> http://pandora.rz.uni-osnabrueck.de/cgi-bin/maketable.cgi (my statistics
> script,
> work in progress)
>
> I found that these problems are still pending:
>
> 1. make install ./. make install-strip
>
> On RedHat, the binaries vary from ~10 MB (install) to ~600K (install-stripped).
> Installation of stripped binaries should be the default, otherwise,
> 10 index/searchd processes will slow down a big box ...
>
> 2. once selected/checked - ever selected/checked
>
> I did not succeed in holding options checked between
> subsequent calls of s.cgi. I uncommented the
> s.htm-dist template lines for the date selection. But this does not work,
> though suggested by the authors. Maybe somebody can drop a line
> how this can be solved.
>
> 3. not indexed yet (status 0)
>
> After successful indexing (not interrupted, everything runs fine),
> there are still URLs with status 0, menas 'Not indexed yet'. Why do they
> appear,
> if index has finished successfully. What can I do to make them disappear?
> There is a thread, dealing with this topic and it was suggested to
> use 'Follow no'. For me, that makes no sense, as the indexer should
> index everything before stopping. Should I use 'index -q ...'.
>
Possibly, "index" couldn't connect to host, where status of some URLs remained
0.
Try to run "index" again.
>
> 4. phrase search ./. stopwords
>
> Searching for a phrase which contains a stopword, the results
> are not satisfactory. It seems as if phrase search returns simple word
> search, combined with AND.
This is not the truth, stopword inside the phrase is assumed to be any word.
If your phrase contains one non-stop word and one stopword, then this is
equivalent just to one word.
>
>
> 5. Order of 'xx -yy' and '-yy xx' should not be relevant, but it is ...
>
> Using a false order, means specifying first -words and then normal
> words, the search is not successful. ('Empty query' message).
>
> 6. 'UtfStorage yes' and consequences ... [URGENT!]
>
> My old database from 1.2.4 was updated to 1.2.5 format
> when using 'index -b' (see below). The new config directive 'UtfStorage yes'
> for aspseek.conf and searchd.conf was suggested and the new
> column 'WordurlU' should be renamed to the still existing name
> 'WordUrl' in the aspseek12 database.
> This sounds strange to mee, as there is no further explanation.
>
> Output of index -b:
> [wwwrun@pandora etc]$ ../sbin/index -b
> Loading configuration from /usr/local/aspseek/etc/db.conf
> Loading charset [iso88591] from table charsets/iso88591
> Loading configuration from /usr/local/aspseek/etc/charsets.conf
> Loading unicode charset [iso-8859-1] from table tables/iso8859-1.txt
> Loaded 283 ngrams from /usr/local/aspseek/etc/langmap/german-iso8859-1.lm
> Loading unicode charset [iso-8859-1] from table tables/iso8859-1.txt
> Loading configuration from /usr/local/aspseek/etc/ucharset.conf
> Loading stop words from /usr/local/aspseek/etc/stopwords/german
> Loading configuration from /usr/local/aspseek/etc/stopwords.conf
> Loading configuration from /usr/local/aspseek/etc/server.conf
> Loading configuration from /usr/local/aspseek/etc/aspseek.conf
> Error: Unknown table 'wordurlu' in <drop table wordurlu>
> 728000 records of 728323 total inserted
> Creating index
>
> Table "wordurlu" with UTF8 encoded field "word" has been created
> Rename this table to "wordurl" and put command "UtfStorage yes"
> to "searchd.conf" and "aspseek.conf"
> Segmentation fault (core dumped)
>
> Which MySQL monitor (mysql) command accomplishes the renaming from
> wordurlu to wordurl? I Used:
>
> mysql> drop table wordurl;
> Query OK, 0 rows affected (0.03 sec)
>
> mysql> alter table wordurlu rename wordurl;
> Query OK, 0 rows affected (0.00 sec)
You should run "rename table wordurlu to wordurl".
Check if table "wordurl" exists.
This is not implemented in "index" intentionally, so you could backup your
"wordurl" table before.
>
>
> But searchd will not start ... (<defunct>).
>
> 7. Ranking algorithm
>
> There is no high level description of the ranking algorithm.
>
> 8. Some parameters of s.cgi cannot be used in s.htm (?)
>
> There is no cross reference to the s.htm description for the following
> s.cgi parameters:
>
> ln=url - shows all links to a given url (referring urls)
> o=# - switches between output templates, no example given,
> which names should be used (res-top, res-bottom, res? for first,
> names for second, third template ?)
>
The same names should be used.
>
> 9. ch=url does not work for cached PDF files
>
> Clicking on the cache link for PDF files [Cached], my server yields
> an 'internal server error'. What mime types must be configured
> to output a cached PDF file (is plain text then) appropriately?
>
> 10. M2 and moreurls
>
> By chance, I found that $M2 refers to 'More from this site'. Is there a $M
> or $M1, too?
2 in $M2 means, that this template should be printed if more than 2 documents
found on this site.
If you will use $M1, then this template will be printed if more than 1
document found on site.
>
>
> ----
>
> I know these are lots of question, but I think it may be welcome to
> 'brainstorm' first impressions with a new release.
>
> Any help and comments are welcome. I will do my part, too ;:)
>
> Frank
> #-------------------------------------------------------#
> Dipl.-Math. Frank Elsner
> Universitaet Osnabrueck (University of Osnabrueck)
> - Rechenzentrum - (Computing Center)
> Albrechstrasse 28, AVZ
> D-49076 Osnabrueck
> Deutschland (Germany)
>
> Tel. (Phone): ++49 (0)541/969-2343 Fax: -2470
> E-Mail: [EMAIL PROTECTED]
> #-------------------------------------------------------#
