Re: ANNOUNCE: Beagle 0.2.15

 You must have already heard about the new release, beagle-
A _lot_ of changes went into this release (0.2.15). I am listing below
the changes directly visible to end users. For the curious amoung you,
I will also mention the revision number. To get the exact changeset,
go to<revision number>&view=rev

Detail Changes

* #3217 (beagle-search): Now you can select the labels in the details pane.
* #3139 (evolution mail backend): Evolution mail backend now stores
the bookkeeping information in a sqlite database, uses much less
memory, specially for evolution with large number of folders or
* #3144 (mp3 filter): If the same tag is present in both id3v1 and
id3v2, then only add the tag once.
* #3157 (evolution backends): Correctly detect incompatible
evolution-sharp libraries. A lot of "mono crashed" kind of bugs were
due to this.
* #3192 (beagled/index-helper): Also check
/proc/acpi/ac_adapter/ACAD/state for power/battery status.
* #3208,3247 (external-filters): Add the sample filters to the shipped
external-filters.xml file. Do not overwrite any modified
external-filters.xml file on the system, but install the sample file
as external-filter.xml.sample.
* #3212 (beagle-dump-index): New tool to figure out everything
(mostly) in the beagle index about a file in your home directory.
* #3213 (index-helper): Index-helper does the actual job of extracting
metadata and text data from files. It does not monitor screensaver to
speed up when user is idle. Only beagled monitors screensaver to speed
up crawling and scanning for data when user is idle. This was broken
in #2720 and fixed in this rev.
* #3224 (kmail backend): Better test to determine if a directory is a
valid kmail folder.
* #3229 (beagle-search): Display the tags and comments of an image in
the details pane. Waste less space and beautification.
* #3231 (logging): New log level - Always. These logging messages are
informational in nature and (surprise) always printed.
* #3243 (rtf filter): New RTF filter based on the RTF parser of Paul
DuBois. Much better than the previous one in all respect.
* #3244 (filters): Some technical tips for writing filters -
* #3249,3274,3304,3323 (filters): Filters return the text data of a
file when asked by the lucene index writer. It can do so in two ways;
1: extract all data in a huge string and then return the string when
demanded; 2: when asked, go and read some data and return it, do this
repeatedly. While the first approach is faster, if performs badly in
terms of memory. The second approach is better as long as the amount
of data extracted each time is not too small. These changes modify the
pdf, html and text filters to follow the second approach.
* #3259 (KOrganizer backend): Except KNode, all of the kontact suite
are covered. Anyone uses KNode ?
* #3264-3266 (Konversation backend): New style Konversation backend
which split the log into daily or hourly sessions. This will cause
much less load on the system when a new line is added to the log file.
This should serve as a good template for other irc chat backends.
* #3280-3281 (scheduler): Add support for Linux's relatively new
SCHED_BATCH scheduler policy. Set the priorities and nice value so
that beagle does not overpower other important CPU tasks.
* #3286-3289 (image filters): Beagle uses F-Spot image parsers to
extract metadata from jpeg, png, tiff and other image files. This
change updates the local F-Spot files; so all the fixes that went into
F-Spot are available. Moreover, Iptc metadata is now extracted from
jpeg files.
* #3290 (word filter): Previously index-helper used to invoke
functions in libwv (word parser library) to filter word doc files. Due
to some bugs in libwv, it sometimes crashed while parsing, crashing
beagle along with it and causing much more trouble. This change moves
the word parsing to beagle-doc-extractor, an external program based on
libwv, so that libwv crashes can not crash beagle anymore.
* #3290 (filters): A lot of the filters depend on external tools to
extract data like pdfinfo, pdftotext, mplayer etc. Sometimes these
tools go crazy on some files (mostly invalid or corrupted) and takes
100% CPU for long time or eats huge memory. Now limits are set on the
maximum time or memory these tools can take. Prevents alien invasion
through bad files.
* #3296 (filters): Filters register what types of files they can
handle using something called FilterFlavor. Previously, anytime a new
filter was created, new filterflavor objects were also created which
is totally unnecessary. Now, creating new filters does not create any
filterflavors. If you have any homemade filters, see the changeset
carefully: you may not call AddSupportedFlavor() in the constructor
and must move them to the overriden RegisterSupportedTypes().
* #3305 (thunderbird backend): Disabled. Use ./configure
--enable-thunderbird in (0.2.15 has a bug) to enable it. Do
not complain if the roof crashes after doing so.
* #3306 (beagle-search): Add a tooltip to the notification icon.
* #3307 (mail filter): Enable searching of name, extension or parts of
name of email attachments.
* #3309 (evo mail backend): No more problem in opening email results
on machines with case-sensitive hostname.
* #3318 (file backend): Beagle stores some indexing related
bookkeeping information in extended attributes; if somehow it cannot
write to xattr (e.g. no write permission), it stores them in a sqlite
database. Using sqlite for storing information is costly, so whenever
beagle gets a chance, e.g. while crawling the next time, it tries to
move the information from sqlite to xattr. This feature was
implemented only partially causing endless looping on a directory in
some cases. This change fixes it. Furthermore, if adds the environment
variable BEAGLE_DEBUG_FSQ which, if set, will log extensive debug
* Many more behind the scenes optimization. Lots of localization fixes
in beagle-search.

Enjoy searching,
- dBera

Debajyoti Bera @
beagle / KDE fan
Mandriva / Inspiron-1100 user

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]