RELEASE-NOTES

---------------
        v.1.2.15
Added option to dirty reconnect, discontinued in MySQL5
This solved bug with lost connection from SEARCHD

---------------
	v.1.2.13
This version can be compiled with GCC up to version 3.3.6
Some minor changes in database connection and tested with MySQL5.1

---------------
	v.1.2.10

This is mostly bug-fixing release. One nasty bug that was introduced in
previous release and resulted in index coredump while doing reverse citations
merging is fixed. Note that you need to run index -H to recreate citation
index if you have used 1.2.9; alternative way is clear the whole database
and reindex everything from scratch.

Several less important bugs were fixed as well. Configuration files were
revised, and some ancient unused stuff cleared out of them. Several minor
documentation bugs were fixed. A few minor fixes went to init.d/aspseek,
aspseek-mysql-postinstall, and .spec file used to create RPM package.

Only one feature was added - DBLibDir directive for searchd and index,
which can be used to add a directory to the list of directories used for
searching database backend library.

---------------
	v.1.2.9

Quite a lot of changes. Several bugs were fixed, including two rare memleaks
in searchd and several coredumps, thus lead to improved stability. This
release should also compile cleanly on FreeBSD.

This release also contains several fixes from Matt Sullivan
<matt@sullivan.gen.nz>. Below is description of patches from the author:

* Fixed non thread safe use of scanner typeTable which caused corruption
  of the table in medium to high load situations (in particular this
  permanently broke use of "-" in queries (until next searchd restart)
  i.e. 'abc -xyz' would become 'abc AND xyz').

* Fixed a small bug in templates.cpp which caused newlines to be added
  before ending font tag during cached page hilighting (effect was that
  cached page would not appear exactly as original in some cases).

* Fixed rare segfault resulting from buffer overflow when creating query
  key for query cache (*many* stemmed words could overflow buffer).

* Improved tag parsing to handle omitted quotes, fixes cases such as
  <A HREF="http://www.server.com"; TARGET=_new">  Side effect is more URLs
  are discovered.  Previously remainder of document would be ignored
  (resulting in URLs were not added).

* Fixed problem where script name has no suffix (exacerbated by addition
  of host to script_name) also removes prepending of hostname to script_name
  (not removed in mod_aspseek although it should be optional here also).

* Initial URL insertion (via "Server" config parameters or use of '-i') and
  URL deletion ('-C'; was this by design?) does not use delmap.

* Fixes order of logging of "Adding URL" in single threaded mode to be
  consistant with both realtime and threaded index modes i.e. log before
  call to HTTPGetUrlAndStore() rather than after (in past I think this
  has been a source of some user confusion when messages such as
  "URL deleted" appear before rather than after "Adding URL".

* Adds support for HTTP method POST to s.cgi.

* Adds feature which allows non-incrementing of hops value when redirects
  encountered. Adds two config options: IncrementHopsOnRedirect and
  RedirectLoopLimit.

Great work, Matt!

Since this version index implements new strategy of indexing "dead" sites
(sites that does not respond to requests). Now number of threads that are
processing such sites are limited to quarter of total number of threads,
as long as there are enough non-dead sites to process.

Also, a new nice feature was added. The "Replace" aspseek.conf directive
now works as sed's "s" command and so can accept \( and \) constructions
in search expression, and \1 to \9 - in replacement. See aspseek.conf(5)
man page for more details. Code was contributed by Jeff Watts
<jeff.watts@ni.com>.

If you are upgrading from 1.2.8 or earlier versions, please note the
following:

1). If you had many sites and reindexed them, it is advisable to run
index -H to re-create citation index files. Versions of ASPseek prior to
this had a bug that caused extra bytes to be written to the above mentioned
files in the process of merging of direct citation index.

2). If you have used "Cache" feature, please rename the SQL table "cache"
to "rescache". This is done with "ALTER TABLE cache RENAME TO rescache"
SQL statement.

---------------
	v.1.2.8

The most interesting new feature in this release is module for Apache
contributed by Ivan Gudym <ignite@ukr.net>. This module can be used
instead of s.cgi frontend. See README.APACHE_MODULE for more details.
Please note that this feature is still experimental.

As a side effect from mod_aspseek development, generic aspseek client C library
is available (s.cgi and mod_aspseek now both use this library). So, it is
much easier now to develop more frontends, please feel free to write one ;)

It was discovered that our code is still need fixing in order to be compiled
by gcc3. This release contains much more gcc3-related fixes, so now we are
able to at least compile it using gcc3 (not sure about how good it will work).

We have also fixed the code to be 64-bit clean and even tested in on Alpha
platform, so people who have such hardware are welcome to do more tests.

This release also fixes problem of s.cgi not working with the new Apache.
Seems that Apache is now closing file descriptor 0 (stdin) before running
s.cgi, and s.cgi code had a bug that was triggered by it.

Also, a major and long-standing bug was fixed. The bug caused "orphaned"
results to be found sometimes, after a few reindexing sessions. "Orphaned"
means results had no content assosiated with it, no title, no cached copy
etc. If you see it on your system, try to use 'index -X1' to check and
'index -X2' to repair your database (these options were introduced in 1.2.6).
Also, if you use 'searchd' from 1.2.8, this will never happen again.

We have received quite a lot of complaints about non-working external
converters in version 1.2.7 and earlier. Most of the people complained
about "Can not get exited child pid" error message printed by index while
running external converter programs. There wasn't an actual error, but
just a message printed by a piece of code we have inserted trying to debug
a problem with dying resolvers; converters actually worked OK. Well, we have
removed this piece of code as it is not needed any more and confuses people.
Another problem we have found it s.cgi was unable to print correct Content-Type
header while showing the "Cached copy" of document that was converted by
external converter. This is fixed now. So external converters are working and
were working before; people who submit bug reports probably have some problems
with configuration or converter programs itself.

One user reported crash when using index. It was discovered that he was set
MaxWordLength to value > 32. So we have implemented a check and limit the
value to the build-in maximum. If you do really want to see words as
large as >32 characters, you need to fix some aspseek code as well as SQL
tables, and only when increase MaxWordLength value in aspseek.conf. So it
is there mostly for decreasing it.

URLs which are marked deleted during crawling are now really deleted after
delta files merging from urlword table and their URL IDs will be reused
during next indexing sessions. This feature helps to decrease maximum
URL ID and thus memory consumed by "searchd" process.

Last but not least, two man pages were added: index(1) and aspseek-sql(5).
Latter page described structure of SQL database used by ASPseek.

---------------
	v.1.2.7

Previous version introduced new CBuddyHeap class; after the release it
was discovereds that implementation was not portable to non-Linux platforms.
More to say, 1.2.6 was uncompilable on some Linux distributions as well.
This release aims to fix that issues. More to say, code should be compilable
by GCC v.3 now.

At last we have found and fixed the reason of "hanging resolvers" bug in
index that appeared in 1.2.5 and hits many people since that time. So people
who is still using 1.2.4a for that reason can now uprgade.

We have also fixed random number generation facility, which apparently was
broken.

Jan Karabina sent a one-line patch that fixes incorrect index misbehaviour
for URLs that are in CheckOnly or CheckOnlyNoMatch (GET was used instead of
HEAD in HTTP query). He also sent us new langmap files for Czech encodings
and corrected Unicode table for cp852 encoding.
---------------
	v.1.2.6

In this version we have fixed numerous small and not-so-small bugs, as well
as added some improvements and new features. Here some changes are described
in details; for full list of changes see the NEWS file.

To summarize, this version should be more stable, as we have found the
reasons and fixed a few coredump cases, as well as healed some memory
leaks. Also, some tricks were incorporated to boost performance: buddy heap
is implemented for storing word position vectors in "index", replacing the
standard STL vector class; CBufferedFile is implemented and is used instead
of stdio during inverted index merging process (also known as "Saving
delta files").

Some other items from NEWS is explained here.

* Added MultipleDBConnections parameter to searchd.conf

Multiple DB connections feature is implemented in searchd. This improves
concurrency between searchd threads, especially in the case when some threads
do pattern search (which is slow), and other thread wants a simple search.

* Exiting of resolver process is logged now
* Event of breaking of read pipe in resolver process is logged now
* SIGCHLD signals are caught by "index"

We have some reports from users of 1.2.5 with complains about stuck of
index process. The problem seems to be in the resolver process. So, we have
added several facilities that can be helpful to nail down the bug.

* Added -X and -H flags to "index"
Use "index -X1" to check inverted index for URLs for which "urlword.deleted"
field is non-zero. Use "index -X2" to fix it by appending information about
deleted keys to the delta files. So if you want to remove records where
"urlword.deleted" is non-zero, run index -X2; index -D, and finally perform
SQL statements to delete unnecessary records.

"index -H" is used to recreate citation indexes and ranks file from
"urlwordsNN.hrefs" fields in case of citation index corruption.

* Fixed -P and -A flags in "index"

"index -P" is used to show "path" to given URL.

"index -A" is used to add or delete site to/from web space.