Issue 101424 - [PT_BR] Update autocorrection for pt-BR (causes Performance Problem)
Summary: [PT_BR] Update autocorrection for pt-BR (causes Performance Problem)
Status: CLOSED FIXED
Alias: None
Product: Internationalization
Classification: Code
Component: localedata (show other issues)
Version: OOO310m9
Hardware: All All
: P3 Trivial (vote)
Target Milestone: ---
Assignee: weko
QA Contact: issues@l10n
URL:
Keywords: performance
Depends on:
Blocks:
 
Reported: 2009-04-29 12:34 UTC by ohallot
Modified: 2013-08-07 15:02 UTC (History)
7 users (show)

See Also:
Issue Type: TASK
Latest Confirmation in: ---
Developer Difficulty: ---


Attachments
New Brazilian Portuguese autocorrection file (77.65 KB, application/x-compressed)
2009-04-29 12:36 UTC, ohallot
no flags Details
Size-reduced autoorrection file for pt-BR (37.12 KB, application/octet-stream)
2009-08-25 19:18 UTC, ohallot
no flags Details

Note You need to log in before you can comment on or make changes to this issue.
Description ohallot 2009-04-29 12:34:00 UTC
Hi,
The Brazilian OpenOffice.org community has improved the autocorrection files
with great improvements over the existing one (in OO release 3.1).

We would like to replace the present autocorrection file with the one attached
below, in the next available release.

Thank you
Comment 1 ohallot 2009-04-29 12:36:15 UTC
Created attachment 61883 [details]
New Brazilian Portuguese autocorrection file
Comment 2 eric.savary 2009-04-29 20:06:51 UTC
Reassigned to SBA
Comment 3 stefan.baltzer 2009-04-30 07:32:23 UTC
SBA->VA: Please proceed, thx.
Comment 4 weko 2009-08-11 14:41:06 UTC
.
Comment 5 weko 2009-08-11 14:59:35 UTC
@ohallot: Could you please check the new AutoCorrection file? The file size of the 
old one is 3KB and the new one's size is *81KB*. If I use the new file in OOo 3.1, 
the Office hangs when I choose pt-BR in the AutoCorrection dialog.
Comment 6 ohallot 2009-08-11 16:13:39 UTC
@va:

I tested on OO 3.1, under Kubuntu with Sun-made OpenOffice.org  and it works.
However, since the file is much larger, it looks like it takes more time to
fully load the content in the autocorrection options.

I placed the (unziped) file in 
/opt/openoffice.org/basis3.1/share/autocorr/

At the end, it works perfectly. 

As a side note, the file is much bigger because since 2009, the Portuguese
language underwent a spelling reform that had modified much of compounded words
(with a dash/hyphen, like auto-correção, which now is autocorreção) and Hunspell
does not handle that. We have not received complains about this new
autocorrection file.

Comment 7 weko 2009-08-14 16:01:35 UTC
@ohallot: I tested the new AutoCorrection on 2 computers (PC & Mac) and it took 
20 seconds up to 55 seconds until the dialog was ready for editing. This is IMO 
not acceptable.

In addition I do not think that the AutoCorrection is the right place to take the 
spelling reform into account. That would be the Brazilian Portuguese Dictionary 
which is actually maintained by you.[1] So such a bloated AutoCorrection is 
superfluous.

@sba: What is your opinion on this matter?

[1] http://extensions.services.openoffice.org/project/Vero
Comment 8 ohallot 2009-08-14 23:22:30 UTC
I beg to differ.

1.This update is available for download in the OOo (pt-BR) community [1] site
since March 2009, and absolutely no complains came to us so far.

2.Besides the delay annoyance, there is no other flaw that can be assigned to
the Autocorrection file  (hangs, errors or anything such), at our knowledge.

3.The file is addressed to the portuguese-spoken users, so it should not concern
others OOo L10n users.

4.As our language have many compound words, Hunspell cannot verify compound
words (with hyphens). That has generated noise in the press at a critical moment
when the portuguese language is changing its spelling rules. 

5.Admitedly, most of the entries drop off in a future Autocorrection update, or,
as soon as Hunspell can address compound words.

6.VERO (the portuguese spelling dictionary for Brazil) got a lot of press
attention because it was the first and prompt response to the new spelling rules
for Portuguese and is way ahead of the same offer from Microsoft which has not
yet set a deadline for the release of their reformed spelling dictionary.

Indeed, the Autocorrection is suited for a situation where the spelling rules
are new to the OOo public, and is handy for fast ramp-up in learning the new rules.

[1] http://www.broffice.org/verortografico/baixar
Comment 9 weko 2009-08-18 12:04:56 UTC
@tl: The updated AutoCorrection (pt-BR) causes an extreme long delay in the 
AutoCorrection dialog when pt-BR is selected. For further information see the 
comments in this issue. - Is there a chance to shorten the delay time? IMO 20 to 
55 seconds are not acceptable.
Comment 10 thomas.lange 2009-08-18 12:15:26 UTC
TL->OS: Since it is about autocorrection please take over. Thanks!
Comment 11 stefan.baltzer 2009-08-24 16:12:32 UTC
SBA-ohallot: about [4] and [5]
Please note that issue 64400 ("dash/hyphen (-) should be forwarded to
spellchecker") is fixed and will be integrated into DEV300_m55 that will be
available very soon. Therefore I suggest the "clean" spellchecker-solution to
address the reform.

The concept of AutoCorrection is "to correct likely occurring errors". A good
example are twisted letters (In English "teh" instead of "the") or triple
letters instead of double ones when fast typing.
 
When the AutoCorrection feature is "hi-jacked" in this way, the user may easily
keep his/her "old" writing habits and "the AutoCorrection" will fix all
wrong-doing".
Since the correction takes place immediately without visible notification, there
is hardly a "teaching effect". So I do not regard this as a good solution "to
teach new rules". 

To me this looks merely like a quick solution for people who have to produce
correct text without having time (or mood or else) to learn. Of course no one
complains when his laziness is supported as this is basically the purpose of
computer use :-).

Szenario_1:
Guess this is included as it is now and then later removed (because a future
spell checker addresses this) then there will be outcries "Bad bug in OOo! I
have to write correctly now. I am used to get all corrected" I believe that this
scenario shall be limited to those who pro-actively install this and not for all
pt_BR users.

Szenario_2: A user who wants to add "real new AutoCorrection data" for his
private or professional needs, thus needs to open that dialog several times per
session. 

Yet-anoter-reason to re-think:
In order to enhance the office performance, are working very hard on several
edges of the office. I am talking about man years of professional full-time
developers that are dedicated to do this.

Putting TL on CC.
Comment 12 ohallot 2009-08-25 19:16:00 UTC
Hi to all

Thank you for the relevant debate and information.

Based on the fact that hunspell will handle compound words in the next release,
I am attaching a new autocorrection file with compound words removed, which
should load in much less time, and hopefully meet the timing annoyance.

Regards
Olivier
Comment 13 ohallot 2009-08-25 19:18:03 UTC
Created attachment 64380 [details]
Size-reduced autoorrection file for pt-BR
Comment 14 weko 2009-08-31 16:44:13 UTC
@ohallot: Okay, with the current AutoCorrection file it takes ca. 20 sec. instead of 
55 sec. So we will integrate it.
Comment 15 weko 2009-08-31 16:57:06 UTC
Reassigned issue to VA.
Comment 16 weko 2009-08-31 16:57:57 UTC
Fixed.
Comment 17 weko 2009-09-01 15:47:54 UTC
Verified in CWS.
Comment 18 weko 2009-10-06 11:40:26 UTC
Verified in DEV300m60 and closed this issue.