Senga is a development group focused on information retrieval software. The primary purpose of the components distributed on Senga is to build a large scale internet search engine. Each component may be used separatly, for instance the full text indexing library is a standalone package that could be used in a text editor or a mail client.

unac unac-1.7.0, Text-Unaccent-1.07 are available 2 September 2002

Link The Unicode-3.2 bug preventing translation of some characters such as æ (ae collated) was fixed by Rune Nordbøe Skillingstad. The perl module was also updated accordingly. A debug facility was implemented to ease the diagnostic and fix of similar problems in the future. Debian packages and rpm are available for the C library (unac) and the perl module (Text-Unaccent).

unac unac-1.6.0, Text-Unaccent-1.06, phpUnac-1.6.0 are available 6 July 2002

Link Unac was upgraded from Unicode-3.0.1 to Unicode-3.2. Perl and php modules were also updated accordingly. Debian packages and rpm are available for the C library (unac) and the perl module (Text-Unaccent).

webbase webbase-5.17 is available 10 September 2001

Link The home_re table map domain names to regular expressions that extract the unique server name. Rules for large free hosting sites need this. Signal handling so that the crawl of a URL is never interrupted. The exploration will stop after or before loading a URL only. -noheuristics works with -touch to force loading even if the content of the URL is in the database. Fix index updating bug that removed documents from the index when they are found Not Modified by the crawler. Upgrade md5 code to GPL and other small utilities fix. Purge unused test data.

mifluz mifluz-0.23 is available 23 July 2001

Link Bug fixes in mifluzsearch restricted queries (-l and -h) and more regression tests. Manual page for mifluzsearch. Synchronize with unac-1.5.0. Use AM_ICONV macro to prevent compilation warnings. Minor distribution fixes.

unac unac-1.5.0, Text-Unaccent-1.05, phpUnac-1.5.0 are available 19 July 2001

Link In this minor release the C library, perl module, php module were updated together. Debian packages and rpm are available for the C library (unac) and the perl module (Text-Unaccent). Better detection of the iconv library, using the AM_ICONV macro of Bruno Haible. Upgrade autotools files. Minor documentation upgrades.

unac Debian package for Text-Unaccent-1.04 is available 18 July 2001

Link The Debian packages for the perl interface of unac-1.4 are available. You will find them at http://www.senga.org/download/unac/debian/. At some point, when and if I become a Debian developer, it will be available on the Debian machines.

uri uri-2.13 is available 16 July 2001

Link The bugs reported were all fixed. They were not many, fortunately ;-) A + in the query part is left untouched to allow +%2B+ to have the expected behaviour. MD5 RSA code replaced by GPL code. Remove include config.h that was wrongly added to uri.h and caused problems when used with other packages.

unac Debian package for unac-1.4 is available 12 July 2001

Link The Debian packages for unac-1.4 are available, thanks to Rémi Perrot (remi_perrot@users.sourceforge.net). To be honest Rémi prepared them long ago but I did not put them online.

catalog Catalog-1.03 is available 11 July 2001

Link This is a maintainance releases that contains many bug fixes and some feature enhancements, in the dmoz handling for instance. symbolic links now have a name instead of borrowing the name of the category they point to. convert_dmoz handles UTF-8 dumps, the whole procedure is simplified. New methods (treecbrowse and treecedit) that display the whole tree structure for people who have catalogs that can fit on a single page, provided by Mejai Maher (mejai@zehc.net) Bug fixes and sqledit enhancements provided by Benjamin Drieu (drieu@bocal.cs.univ-paris8.fr) and Takanori Ugai (ugai@flab.fujitsu.co.jp) gif files translated to png Fix the problem of displaying ISO-8859-1 accented chars in category names

uri uri-2.12 is available 9 July 2001

Link The uri library was moved from SourceForge to Savannah. There are many reasons for this move, the most proeminent being that Savannah is developped cooperatively and we can contribute to its evolution and maintainance. Another reason is that the Free Software movement has a philosophy that we like. Internationalization was added by Florian Hatat (mininet atwanadoo.fr) and messages in French are available.

mifluz mifluz-0.22 is available 2 July 2001

Link Implement result cache for mifluzsearch. The results are stored in /var/cache/mifluz for a given amount of time and re-used when possible. If the search only asked for 10 documents, only those are cached. When the search asks for more, the cache is filled accordingly. The cache is a single Berkeley DB file with subdatabases. It can be dumped using htdb_dump. Fix mifluzsearch bug that prevented searching indexes where the document did not start immediately after the word. Fix mifluzsearch bug that ignored mandatory words that do not exist in the index. Synchronize with unac-1.4.0 Removed obsolete benchmark results

Senga Web redesign in progress 28 June 2001

Link After too long I'm back on Senga. Software evolved (almost all of them) although I did not take time to make releases. Thanks to Igor Genibel we now have Debian packages for about everything. I spent a huge amount of time building a development infrastructure that Senga could use (Savannah). I spent approximately the same time founding FSF France and FSF Europe, mainly because it's pretty useless to write Free Software if you don't worry a bit about its future :-)