Breaking News!... Help pages can be edited directly in gerrit.

l10n
#1

Hi L10N Community! Breaking news!

Help pages can be edited directly in gerrit, all editions done in your
browser.

This means that edition of textual XML is possible with automatic
generation of a patch submitted to gerrit. Therefore, English source
help files can be edited or corrected directly in master and with
security and peer review provided by TDF security infra. Such
capability will be extremely handy for ad-hoc correction of typos and
linguistic mistakes we often do in English, as the LibreOffice project
is developed by many non-English native speakers.

This capacity will let help writers a direct access to patching without
having to download and build LibreOffice and its help. Although the Help
XML knowledge is still need to correctly write a help page, some of our
skilled NL leaders and translators can now fix typos directly, without
passing by the lengthy process of reporting typos, building help and more.

Please see a tutorial on editing help source files in our wiki page:

https://wiki.documentfoundation.org/Documentation/GerritEditing

Happy help page fixing !

Cheers

#2

Great news, Olivier, Thanks!

I hope to be able to use it often :slight_smile:

Cheers,
Cor

#3

Hi all,

Hi L10N Community! Breaking news!

Help pages can be edited directly in gerrit, all editions done in your
browser.

This means that edition of textual XML is possible with automatic
generation of a patch submitted to gerrit. Therefore, English source
help files can be edited or corrected directly in master and with
...
Please see a tutorial on editing help source files in our wiki page:

https://wiki.documentfoundation.org/Documentation/GerritEditing

Great news, Olivier, Thanks!

I hope to be able to use it often :slight_smile:

Be really careful however that each change will go through each
language, so be sure that what is committed is the right fix/enhancement
and that there will be no other change to the string.
Cheers
Sophie

#4

Important warning - thanks :slight_smile:

I've read the Wiki swiftly, and it is clear that - though convenient -
it's indeed not to use without care.

#5

These are great news! Love it!

Exactly what I was asking about a day before in one of our IRC channels,
b/c I was clueless about how to use gerrit properly.

I uses the opportunity to commit my first patches (only whitespace cleanups for now)
to gerrit in time of our Hackfest / German Community Meeting in Hamburg.
So yes, it works! :smiley:

What would be really really good were a) a review of course and b) an jenkins/ci buildbot for help
for mistakenly merge patches which break master or branches in cases of backport.

#6

Hi,

And now the Babilon tower is open to all ...

For years I was asking for a kind of a review process for all new strings
in UI/help and for the editing process. So someone overlooks all the
changes going in and stops the ones that seem unfit (in langague or
technical sense), thus reducing the repetitive editing of same strings.

Instead, now there is a possibility to edit everything even more "easily",
with even less thinking about it before it is done.

So now we might see a surge in help changes that make many l10n projects
quite vulnerable, and most probably several edits of same strings in a row
(a documenter reminding oneself that another thing should be added or
changed or edited, after it is already featured in git).

Hopefully, this function is limited to master, because if this is available
in "stable" versions, this will lead to havoc ...

Lp, m.

#7

Hi,

just one example from Gerrit 4 hours ago:
"vector graphics;converting bitmaps"
was changed to:
"vector graphics; converting bitmaps"
See:
https://gerrit.libreoffice.org/#/c/52546/2/source/text/simpress/guide/vectorize.xhp

Is that "missing space" an error? Not. What does this change cause? It
causes 100+ translations of that string obsolete, fuzzy, untranslated, not
shown in builds ... And 100+ localizers must again localize, check, edit
this same string ...

Is this kind of editing/"cleaning-up" of help really useful/desired? Not.

Lp, m.

#8

Hello Sophia

Glad you and others liked it.

The Foundation Manifesto includes "to an open and transparent
peer-reviewed software development process where technical excellence is
valued"

TDF will not spare efforts to appreciate all contributions by the
community. Lowering barriers to code contributions is critical to the
Foundation mission.

And thanks for the patches. Much needed. All of them being reviewed and
tested by committers.

Kind regards
Olivier

#9

Hi,

it causes visual glitches, like comma faults, at least in the old help system.

So yes, I think so.

Assuming the most translation were done already correctly
it is also an opportunity to review own translations.

Otherwise there are just 2 clicks, no?

And it is not a long where it may "disturb" translators.
Once it is done, it is done. Except new strings of course.

And there don't come in a big bunch and I am also testing in the moment,
which are the best practices to get rid of some failures
and to make our help system and files more consistent.
I have no problems with abandoning my patches for good reasons.

It is a learning process for me too.

#10

Hi, Sophia,

I do not see these glitches in LibreOffice help. Can you send me a
screenshot?

Thanks, m.

#11

I experienced them in older versions of LibreOffice in German, where the old behaviour were used.

I changed (and proofreaded, while at it) already our help in German with the syntax:

<bookmark_value>blah; blubb</bookmark_value><boookmark_value>foo; bar</bookmark_value>

So there should be correct.

But I will have a closer look in the next days for examples.
I think the patches with this kind of changes in
can wait and lurk in gerrit meanwhile. :wink:

Others can be merge meanwhile,
b/c the double spaces appear in flowing text only.

#12

Well, the double spaces problem ...
I have removed all occurences in the Slovenian help, and now I will be
forced to go through that again manually ... L10n teams/languages do not
have same errors, see, and this way we are spending time on things that we
might have fixed long ago ...

Besides the fuzzy mark gettext should obviously be extended with a
"source-string-changed-but-it-might-not-affect-your-existing-translation-because-it-was-a-minor-edit"
tag which should not lead to untranslated status of the string in code and
the existing translation should be pulled from git as still a valid one.

This tag could be called in short "check-translation" or
"minor-source-change" or something similar.

Is this feasible with the gettext developers?

Lp, m.

#13

This were the best solution, yes.

#14

Hi,

Hello Sophia

Glad you and others liked it.

The Foundation Manifesto includes "to an open and transparent
peer-reviewed software development process where technical excellence is
valued"

TDF will not spare efforts to appreciate all contributions by the
community. Lowering barriers to code contributions is critical to the
Foundation mission.

Breaking others work like builds and translation workflow is not what
exactly TDF values I hope...

And thanks for the patches. Much needed. All of them being reviewed and
tested by committers.

Patches adding spaces in a string for an index? Again I don't see how
adding a lot of work for contributors will help them to contribute. There
are a lot of typo in the new help content that should be corrected before
and not just pushed as is, expecting to be corrected by translators. Asking
more and more to them will just push them away as it's already the case for
several languages. Sorry but I'm really not happy by this dangerous move,
consequences are too high for the moment.
Cheers
Sophie

#15

Hi Sophia,

I experienced them in older versions of LibreOffice in German, where the
old behaviour were used.

I changed (and proofreaded, while at it) already our help in German with
the syntax:

<bookmark_value>blah; blubb</bookmark_value><boookmark_value>foo;
bar</bookmark_value>

So there should be correct.

I don't understand what you want to achieve here. The space between
'blah;' and 'blubb' is not needed, the help index is build as blah=first
level, blubb=second level of indentation.
See here for an explanation
https://wiki.documentfoundation.org/UI_and_Help_files_Content_Guide#.3Cbookmark_value.3E.3C.2Fbookmark_value.3E
I won't add any space because as I said it's a great way to recognize
that it's an index entry.

But I will have a closer look in the next days for examples.
I think the patches with this kind of changes in
can wait and lurk in gerrit meanwhile. :wink:

Others can be merge meanwhile,
b/c the double spaces appear in flowing text only.

There are a lot of double spaces added in sentences where it is real
spaces (and will have an incidence in the look and quality of the en_US
help), I think (I might be wrong) this is the a problem with the help
tool that should be corrected, or at least be removed before committing
the patch. I've absolutely no time to dedicate to it at the moment, but
if someone would like to have a look at it, it would be great.

Cheers
Sophie

#16

Hi,

Regarding adding the space between the first and the second index entry inside the bookmarks,
I think it would still be okay to change it to but would cause too much noice for too little win (consistency), indeed.
So no need to spend more efforts in this. All I wanted to know. :wink:
It was just a random file I picked and submitted to gerrit to get a feedback about this kind of changes.
I did it also in other patches wherein I will revert that.

And I will also fix the other issues I caused in my patches regarding XML syntax of course,
hoping the xmllint checker chloph wants to integrate will help me and others in this regards
so nobody needs to review XML mess in help related commits anymore.

#17

Hi Sophia, *,

[…]
And I will also fix the other issues I caused in my patches regarding XML
syntax of course,
hoping the xmllint checker chloph wants to integrate will help me and others
in this regards
so nobody needs to review XML mess in help related commits anymore.

FYI: that is active now - it doesn't build LibreOffice, it just uses
xmllint to check the xhp files against the dtd, and thus should be
pretty quick.

ciao
Christian

#18

Hi cloph, *,

many, many thanks, very much appreciated. :smiley:

Btw: Someone claimed a misspelled word in German UI: "Zahl~format" instead of "Zahlen~format",
how it were right.

How can I find and fix such strings? Opengrok and Gerrit? Pootle wasn't helpful at all in this regard.

And is it possible (or even helpful) to enable xmllint for changes in .xc* configuration files?

#19

Hi Sophia

Hi cloph, *,

many, many thanks, very much appreciated. :smiley:

Btw: Someone claimed a misspelled word in German UI: "Zahl~format"
instead of "Zahlen~format",
how it were right.

How can I find and fix such strings? Opengrok and Gerrit? Pootle wasn't
helpful at all in this regard.

If you fix it in Gerrit, it will be overwritten in Pootle. You can
search for it in Opengrok or Gerrit, then correct it in Pootle. You can
also install the .qtz package to get the KeyID of the string
see here how to install it:
https://wiki.documentfoundation.org/Language

Cheers
Sophie

#20

Hi Sophie, *,

If you fix it in Gerrit, it will be overwritten in Pootle. You can
search for it in Opengrok or Gerrit, then correct it in Pootle. You can
also install the .qtz package to get the KeyID of the string
see here how to install it:
https://wiki.documentfoundation.org/Language

Cheers
Sophie

Exactly my problem: In Pootle is not such occurence of "Zahl~format" in our translation
but a hint as comment: "Format -> Zahl~format".
So I assumed the problem with the wrong written word must be somewhere else,
most probably in my poor knowledge in some .xc* file not covered by Pootle.