- redesign of ordering structures in kelondro (old did not work with strict generics)
- 50% IO reduction during read access on kelondroFlex (ommiting of read on index table)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4320 6c8d7289-2bf4-0310-a012-ef5d649a1542
- changed build script to use java 1.5 compiler
- first stept to resolve missing generics definition (about 400 from over 4100 'missing'-warnings)
- added key-iterator to kelondro databases (for rapid from-memory enumerations, will be used for domain name collection, not used yet)
please set your development environment to use java 1.5!
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4292 6c8d7289-2bf4-0310-a012-ef5d649a1542
- the index administration now uses the same code base for url selection and collection
as the search interface. The index administration is therefore a good test environment for
ranking order control
- removed old postsorting-algorithms, will be replaced with new one
- fixed many bugs occurred before during ranking; especially the contraint filtering method
removed too many links
- fixed media search flags; had been attached to too many urls. The effect should be a better
pre-sorting before media load within snippet fetch
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4223 6c8d7289-2bf4-0310-a012-ef5d649a1542
to do this, the index administration interface was extended.
It is now possible to select parts of a index.
See properties shown in interface after a word search for details.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4218 6c8d7289-2bf4-0310-a012-ef5d649a1542
- enhanced searchtime in kelondroRowSets
- enhanced uniq() - reverse enumeration causes less time in case of mass removal of doubles
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4207 6c8d7289-2bf4-0310-a012-ef5d649a1542
the enhancement was made by using better organized data structures and
multi-threading during the sort. A sort can be divided into two separate
processes when the first partition of the quicksort algorithm was done.
Generating a separate thread and starting the thread takes only 10 milliseconds,
so using a separate thread makes only sense if the data amount is large.
statistics about the speed-up:
without ehancement: 250 milliseconds for 100000 entries
with data structure enhancement: 170 milliseconds for 100000 entries
with additional second thread (if second processor is present): 130 milliseconds.
For dual-processor systems, this means about 100% speed-up
a test can be made with the following command:
java -classpath classes de.anomic.kelondro.kelondroRowCollection
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4198 6c8d7289-2bf4-0310-a012-ef5d649a1542
- before absolute paths would be expanded incorrectly, e.g.: fooPath=/a/b/c would become /path/to/yacy/root/a/b/c. Now you can put nearly every dynamically generated data with a configurable path to a location outside of yacys root dir without having to use symlinks (probably good for third party distribution packaging).
- abstractServerSwitch.getConfigPath(setting, default) returns a File instance, either with an absolute path or relative to the applications root path.
- exceptions (hardcoded):
DATA/LOG/yacy.logging
DATA/SETTINGS/httpProxy.conf
DATA/SETTINGS/user.db
TODO: all of these are the global configuration files and they should probably be put into _one_ command line configurable settings path, so it would be possible to package them in /etc/ for example.
- add missing workPath to yacy.init (it was used in code, but there was no default in the file)
- fix broken skinPath (was skinsPath in yacy.init but skinsPath in the code) + a few other broken config reading caused by typos.
- replaced path setting names and their default values with the related static fields in plasmaSwitchboard where not already done/existing
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4196 6c8d7289-2bf4-0310-a012-ef5d649a1542
- Test wCacheMaxChunk against maxURLinCache(), not getMaxWordCount(). This triggered a flush everytime dhtFlushControl() was called.
- If triggered, flush at least 1 entry.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4187 6c8d7289-2bf4-0310-a012-ef5d649a1542
- final flush only when tabletype = RAM
- prestacker (dns prefetch) only if tabletype = RAM and busytime <= 100
- number of maximun entries in stacker is configurable in yacy.init (stacker.slots)
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4186 6c8d7289-2bf4-0310-a012-ef5d649a1542
- removed web structure picture from indexing menu and grouped it together with htcache monitor
- added a database for terminated crawls, when a crawl is finished it is automatically moved to the new database
- extended crawl profile edit servlet, shows now also terminated crawls
- option that was used to delete profiles is now redesigned to a function that moves the current crawl to the terminated crawls and removes all urls from the current queues!
- fixed here and there problems with indexing queues
- enhances indexing speed by changing cache flush sizes.
- changed behaviour of crawl result servlet: the list of crawled urls is shown if there is one, othevise the overview window is shown
attention: the new profile databases are not compatible with the old one. current crawls will be lost! the web index is not touched.
next steps: the database of terminated crawls can be used to start with them a new crawl. This is useful if one wants to re-crawl specific pages and wants to use a old crawl profile.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4113 6c8d7289-2bf4-0310-a012-ef5d649a1542
search profiling showed, that a major amount of time is wasted by computing url hashes. The computation does an intranet-check, which needs a DNS lookup. This caused that each urlhash computation needed 100-200 milliseconds, which caused remote searches to delay at least 1 second more that necessary. The solution to this problem is to attach a URL hash to the URL data structure, because that means that the url hash value can be filled after retrieval of the URL from the database. The redesign of the url/urlhash management caused a major redesign of many parts of the software. Since some parts had been decided to be given up they had been removed during this change to avoid unnecessary maintenance of unused code.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4074 6c8d7289-2bf4-0310-a012-ef5d649a1542
- snippets are not fetched by browser using ajax, they are now fetched internally
- YaCy-internat threads control existence of snippets and sort out bad results
- search results are prepared using SSI includes
- the search result page is visible right after the search request, the results drop in when they are detected
- no more time-out strategy during search processes, results are shifted within queues when they arrive from remote peers
- added result page switching! after the first 10 results, the next page can be retrieved
- number of remote results is updated online on the result page as they drop in
- removed old snippet servelet (which had been also a security leak btw)
- media search is broken now, will be redesigned and fixed in another step
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4071 6c8d7289-2bf4-0310-a012-ef5d649a1542
some urls are fetched so the url cache can be filled with these urls
- the url-prefetch is used to sort out some unresolved urls
- the snippet-fetcher is triggered with the search event id. This is used
to remove missing snippets from the search cache so they will not be displayed again
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4060 6c8d7289-2bf4-0310-a012-ef5d649a1542
it is necessary for the new search process that will do automatic re-searches
a positive effect is, that when a re-search is done it can be monitored how many
results had been contributed from other peers. The message for this contribution
was moved from the end of the result page to the top.
* enhanced re-search time when a global search was done an the local index has
already a great number of results for this word
* re-organised presearch computation; must be further enhanced
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4059 6c8d7289-2bf4-0310-a012-ef5d649a1542
collection objects has been reduced to 50%:
- set new memory calculation functions for indexing process
- adjusted guessed memory amount
-> Testing needed:
try new recommended value (see performanceQueues) and see if OOMs occur.
-> report maximum recommended value, so we can set new default values.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4053 6c8d7289-2bf4-0310-a012-ef5d649a1542
- re-designed remote request result processing
- re-designed local result accumulation, will be further enhanced with snippet fetcher
- removed search process handling in switchboad
- made snippet class static (there is no need for multiple snippet objects)
- removed some redundant tasks in server-side search process, should be a little bit faster now
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4043 6c8d7289-2bf4-0310-a012-ef5d649a1542
- different handling of link quotation
- different handling of link normalization
- enhanced html/unicode en/de-coding
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3993 6c8d7289-2bf4-0310-a012-ef5d649a1542
such an entry cannot be instantiated without allocation of new byte[]; instead
it can re-use memory from other kelondroRow.Entry objects.
during bugfixing also other bugs may have been solved, maybe the INCONSISTENCY problem
could have been solved. One cause can be missing synchronization during bulk storage
when a R/W-path optimization is done. To test this case, the optimization is currently
switched off.
More memory enhancements can be done after this initial change to the allocation scheme.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3536 6c8d7289-2bf4-0310-a012-ef5d649a1542
this shall be used to store a fragment of the index on another physical device,
to split IO load and enhance access speed. The index is splitted in such a way
that the LURLs are stored to the secondary location, and the RWIs to the primary
location. This is especially useful for environments where symbolic links are
not possible and may cause IO access even if there is no write access to the
device which hosts the symbolic link.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3519 6c8d7289-2bf4-0310-a012-ef5d649a1542
The new collection index will be more generalized to support other indexes
i.e. YBR block-rank computation. A clean-up of the many conditions to support
the old database was necessary.
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3506 6c8d7289-2bf4-0310-a012-ef5d649a1542
- better memory allocation for FlexTable indexes
- splitting between static index and dynamic index (only the dynamic part must grow)
- to enable a merge-iteration of new splittet index, a huge number of classes needed to be adopted for new iterator classes
- added new iterator classes that support cloneable iterators
- adopted all iterator classes to implement cloneable itarators
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3453 6c8d7289-2bf4-0310-a012-ef5d649a1542
- each cache can now allocate as much memory as is available
- no more fixed limits
- replaced old performance memory monitor by new one
- added supervision methods as static functions into the classes that provide cache functionality
- steering of ram allocation is done with two simple limits that are ram availability-relative
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3434 6c8d7289-2bf4-0310-a012-ef5d649a1542
too large collection arrays are now avoided. By default, the biggest
collection index is 7. larger collections are dumped into a commons
directory, but cannot yet be used. Bevore doing a dump, the collection
is splittet into a part which has only root-references, and stored back
to the collection; the remaining part goes to commons
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3426 6c8d7289-2bf4-0310-a012-ef5d649a1542
- permanent cache flush is switched off. The optimized cache flush
works better if it is a large number of collections that is flushed
together
- the flush size can be configured instead the flush divisor. There is
only one size for all flushes
- collection records that shall be removed during collection transition
(jump from one collection file to another) are now not really removed
but only marked in RAM. add-operations to the collection use these
marked collection spaces
- index bulk write operations are now separated for each file of a kelondroFlex
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3414 6c8d7289-2bf4-0310-a012-ef5d649a1542
- collections are now hand-over as collection lists to collection index for merge opertations
- collection index lists are separated into 'new' and 'extend' lists
- lists are written separately
- write operations are done into array sets and array indexes. These are now serialized
- write operations into index files are sorted by index;
that means that a R/W head does not need to go forward
and backward, only forward
More enhancements are possible
git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3407 6c8d7289-2bf4-0310-a012-ef5d649a1542
orbiter