Hi Laurent,
i would have expect that is was document-based
If it were document based it would have required an
initialization and deinitialization for that specific
document. And not requiring this seemed to be easier use.
And also if you consider that documents might not be closed
proper for what ever reason or unproper programming (missing
the deinitialization) on the side of the caller, I thought
it to be the better choice to do without and thus eliminting
the chances for memory leaks as well.
And if you consider a server with multiple users it becomes
even more unpredictable.
Really ? the spellchecker code execution is shared throughout users ?
The implementations of the services for the linguistic are all
single-service implementations. That is there will always be only
a single instance of that object. This eliminates the need for
duplicating implementation data. And (depending on the implementation)
buffering a large amount of data obtained from the data files for
each instance. That is of course only if the implementation likes
to do such buffering.
Thus if the same process is used there will be only one instance
as well for several users.
I must admit though that I'm currently not aware if each user gets
spawned a process of its own or if they share the same.
so back to the initial question
how is it possible to underline the words ?
With that is think i could manage to process with a cursor (real time
with a keboard event) - no use of spellchecker
Currently using the API it is only possible to do this by modifying
charcter attributes. But this has the previously described disadvantages.
i'm a bit disapointed though regarding spellchecker implementation.
If it is about spellchecking (i.e. to my understanding a word-by-word
process) you should be fine with implementing a spellchecker and you
need to care about the red wavy lines at all.
But if you like to to do sth sentence based (like proof-reading or
grammar checking) we are currently at a loss. :-(
We are still missing a real solution here.
Anything that can be done about this for the time being must get along
with means of existing UNO API only. And with that at least the problem
with the underlining can not be solved completly to ones satisfaction.
Everything will be just a workaround until we have such functionality
available in the API.
All the other things: iterating through text, identifying word
boundaries, selecting text and replacing it should be fine with
the current API.
Thomas
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
