New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
TextLevelRule conflicts with Java rules usage in LibreOffice plugin #669
Comments
The cache itself isn't used in LibreOffice. It's only used when the
That's because that check only tests a tiny one-word (or so) "sentence" that's so short it doesn't benefit from caching. You can use |
Excellent. Many thanks. Many Java rules seem to already work properly, so maybe it is only the ones that are yet to be fully converted that have misbehave, although the UppercaseSentenceRule was actually converted. Can the issue in it be related with the column adjustement workaround? The rule triggers in the beginning of a paragraph but not inside it, so:
Can the removal of the conditional in this way work without affecting the HTTP behaviour? |
I'm not sure, I guess LO/LT integration works on character positions instead of lines and columns. |
You are right. It doesn't. I tried it now and it does not change, but I am having odd results. If I change the language, e.g. to italian, I get the orthographical error triggers and the sentence trigger. For the untrained eye (like mine) it seemed a much easier solution. I will report any advances, but all tips are welcome. |
Ignore my commit in this thread, this was for issue #699. My bad. |
I set the tokenizeText in the check for paragraph to 'true'. |
Brilliant work, Fred. |
Thank you Tiago,
|
I can confirm this. I tested it with the 'greeting and farewells' rules that I localized from German to Portuguese, since they have the exact same behaviour in both languages. For example, it replaces the dot in I do not know if this behaviour is new. In the coming days, I will test with an older release and see if it also happens, to see if it is somehow related with TextLevelRules.
If someone is working on it it is great. This part of the code links to too many parts, that have to be understood deeply. It would be great if you picked it up, but I know how disheartening spending time with this can be. Regarding these problems, should a new issue be opened? |
The problem with the dialog is not related with the TextLevelRules. I testet it with rules on sentences level.
|
This issue seems solved. |
After the implementation of the new caching mechanism, several issues appeared in the LibreOffice plugin.
The first and most obvious and pressing is the issue with UppercasecaseSentenceStart, that does not trigger properly inside LibreOffice, as seen in the sentence below.
"An error example, of this regression. it is possible to find the error in the beginning of this sentence."
Other related issues are the conflicts with prioritization in Java rules, seen in the Portuguese rules based in AbstractReplace class, and that affect other languages as well.
Last but not least, is the conflict with bitext rules, that demanded it to be disabled in the HTTP server, which may make it unavailable also in off-line usage of some plug-ins.
I would like to work on it, but I do not know where to find the relevant performance data that shows the benefits of the new caching mechanism, so that I can avoid further regressions.
In LanguageTool Regression test pagethe performance graphs do not show any statistically significant difference on any of the performance metrics, so I believe that the metrics that guide this change are not available on that portal.
Where can I find the relevant data? Am I allowed to work on it, or more work on that function is yet to be pushed?
PS - Not that not of the reported issues affect the website, that is working perfectly with all the previous functions.
The text was updated successfully, but these errors were encountered: