Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
ORecordList uidAt(uint index ) added
Speed Improvements at the SQL backend
do not load the list of uids until it's really needed
do not reload the uid list until it's really needed
we got a bitfield m_dirty there...
|
|
Scrolling is now noticeable faster with 10.000 items
but not as fast the XML backend...
OPimCache can be tuned and Query->OTodo too
|
|
I tested it with my todolist and it 'works' for 10.000 todos the hits are awesome ;)
The read ahead functionality does not make sense for XMLs backends because most of the stuff is already in memory. While using readahead on SQL makes things a lot faster....
I still have to fully implement read ahead
This change is bic but sc
|
|
|
|
The SQL-Stuff is currently deactivated. Otherwise everyone who wants to
compile itself would need to install libsqlite, libopiesql...
Therefore, the backend currently uses XML only..
|
|
Case number 4 is still to go
|
|
|
|
|
|
|
|
|
|
The DB layout is not fully what we've in mind
but for example I do lack the Categories infrastructure
for that
I've to implement sorted lists and then I'll make
Todolist fast
|
|
|
|
|
|
|
|
|
|
move from Opie::XMLElement to custom parser
this cuts down loading 10.000 items from 22 seconds
to 4 seconds
|
|
please do not expect that anything is working !
I will debug that stuff in the next time ..
Please read README_COMPILE for compiling !
|
|
|
|
This would allow us todo a paint hack in QTable
|
|
but there is still room for improvisation
|
|
current() for Iterator
setCurrent() for Iterator
count() for both
|
|
|
|
awesome
I had problems with QShared because I forgot to copy it in &operator=
And the other one was d = 0
|
|
|
|
and addition to OTodoAccess
overDue and effectiveTodos
|
|
I had to create templatebase so orecordlist and opimaccesstemplate
do not have circular dependency
otherwise just compile fixes to my templaes
and i did not make them inline
|
|
|
|
I did not know what to write
|
|
|
|
in OPimRecord::setUid
ORecordList implemented never compiled it though
OTodo a new c'tor and some more changes
|
|
|
|
the Records
The CROSS_REFERENCE branch will get ported to it.
We use templates for several reasons
They allow us to share code and to be easily extended
I've to make them not inline to save memory...
I've to port all DBs and Record related classes
|