[Libreoffice] autocorrect limits

[Tommy]

On Wed, 01 Feb 2012 22:37:40 +0100, Michael Meeks <michael.meeks at suse.com>
wrote:

>
> On Wed, 2012-02-01 at 21:03 +0100, Tommy wrote:
>> Hi, as far as I know, the more entries you have inside a OOo/LibO  
>> acor.dat
>> file the longer the replacement table will take to load.
>
> 	That is as expected -but- this should be an incredibly fast operation
> (AFAICS), unless there is some really dumb N^2 or worse in the
> implementation; and there should be no reason for that.


I confirm this being incredibly slow on Windows (don't know how it behaves
on Linux) and has always  been the same from the OOo 2.4 era to current  
LibO 3.4.5

>> actually a 36000 entries .dat file will take 4 minutes to open
>> while with 60000 entries .dat file will take between 15-20 minutes to  
>> load.

you can try yourself with this files:
https://issues.apache.org/ooo/attachment.cgi?id=52416

it's an italian acor.dat file which is almost full (close to the 65K
entries limit)
loading the italian replacement table will take 15-20 minutes
https://issues.apache.org/ooo/show_bug.cgi?id=101726

adding few more autocorrect entries will make the .dat file crash, loose
all data and reset to 0.
https://issues.apache.org/ooo/show_bug.cgi?id=87672


> 	Um - this is not good ;-) It would be good to get a valgrind profile of
> that, or even better just run in gdb and hit ctrl-c a lot as it is going
> - that should show you what is up [ assumign you're compiled with debug
> symbols ].
>
> 	Then apply a little thought, and a bit of coding & life should get much
> better - surely ? I assume you've come here for hacking tips - since
> this is the developers list right ? :-)

well, I'm still far to have any decent coding skill... but maybe one day I  
will have some.
however I'm trying to understand how those issues happen and gather as  
many informations
to start thinking how trying to fix them...

for these reasons I'm monitoring the devs list and when I bumped into  
Kohey's thread and
read about the "reference 16-bit 65535 issue" and how he fixed, I thought  
that a similar
solution could be used for the autocorrect issue as well.

>
> 	All the best,
>
> 		Michael.
>

thank you. I really enjoy the athmosphere of the TDF/LibO group.
here everybody seems open for collaboration and help.
this is a true open-source environment.

Cheers, Tommy