AND-, OR-, XOR-operators in Help

l10n
#1

Hi,

In LibreOffice 5.3 Help on Pootle there are several strings about operators, e.g.:

<link href="text/sbasic/shared/03060100.xhp" name="AND Operator [Runtime]">AND Operator [Runtime]</link>

<link href="text/sbasic/shared/03060600.xhp" name="Xor-Operator [Runtime]">Xor-Operator [Runtime]</link>

Notice that sometimes the logical operators are capitalized and sometimes there's a "-" for joining (which is logical for my language).

Q: Do you know if this is done on purpose, or is this worth notifying the docs-team?

Best regards,
Sveinn í Felli

#2

Hello Sveinn

Makes sense to have it all capitalized.
I'll patch the code
Kind regards
Olivier

#3

Þann mán 28.ágú 2017 12:37, skrifaði Olivier Hallot:

Hello Sveinn

Makes sense to have it all capitalized.
I'll patch the code
Kind regards
Olivier

Thanks Olivier.

Sveinn

#4

… and the fix is now merged for 6.0:
https://cgit.freedesktop.org/libreoffice/help/commit/?id=ba1b065b296a1fce482bb3ebfa2fc8b765b45456

Thanks, Sveinn and Olivier!

Adolfo

#5

Þann þri 29.ágú 2017 23:10, skrifaði Adolfo Jayme Barrientos:

… and the fix is now merged for 6.0:
https://cgit.freedesktop.org/libreoffice/help/commit/?id=ba1b065b296a1fce482bb3ebfa2fc8b765b45456

Thanks to you.

I think there are also some occurrences of 'Or' instead of 'OR':

<https://translations.documentfoundation.org/is/libo_help/translate/sbasic/shared.po#unit=30542403>

Best,
Sveinn

#6

Will this patch render all these string as fuzzy for l10n teams?
Lp, m.

napisala:

#7

The AND-, OR-, XOR-operators are only a handful, anyways I translated mine capitalized as they should have been.
But many strings already became (unnecessarily) fuzzy:

Choose <emph><switchinline select="sys"><caseinline select="MAC">%PRODUCTNAME - Preferences</caseinline><defaultinline>Tools - Options</defaultinline></switchinline> - %PRODUCTNAME - Print</emph>.

has become:

Choose <switchinline select="sys"><caseinline select="MAC"><emph>%PRODUCTNAME - Preferences</emph></caseinline><defaultinline><emph>Tools - Options</emph></defaultinline></switchinline><emph> - $[officename] - Print</emph>.

You see that the enclosing <emph> tag became nested for each string snippet. These are symptoms of a new editor or something different in how the markup is applied - and a very unfriendly burden on translators/proofreaders. The latter is also bloating the codebase.
But are these things avoidable?

Best regards,
Sveinn

Þann mið 30.ágú 2017 07:18, skrifaði Martin Srebotnjak:

#8

So, these changes will make 100+ l10n teams to re-translate/check those
touched strings, yes?
Lp, m.

30. avg. 2017 10:30 AM je oseba "Sveinn í Felli" <sv1@fellsnet.is> napisala:

#9

Þann mið 30.ágú 2017 09:14, skrifaði Martin Srebotnjak:

So, these changes will make 100+ l10n teams to re-translate/check those
touched strings, yes?

Well, these have been brewing for some time; think we have to live with that. But of course we should make noise if there is a bunch of such 'cosmetic' modifications, all at once.
Meanwhile, we could think seriously about why the help/docs markup is so complex (instead of being more CSS-like?) and why there's so little separation of content and formatting - and what can be done???

Best regards,
Sveinn

#10

Þann mið 30.ágú 2017 09:14, skrifaði Martin Srebotnjak:

So, these changes will make 100+ l10n teams to re-translate/check those
touched strings, yes?

Well, these have been brewing for some time; think we have to live with
that. But of course we should make noise if there is a bunch of such
'cosmetic' modifications, all at once.
Meanwhile, we could think seriously about why the help/docs markup is so
complex (instead of being more CSS-like?) and why there's so little
separation of content and formatting - and what can be done???

I would also add why strings have so many tags when the only
translatable content is outside of the tags or in a single place in
the string, like:

<image src="cmd/sc_compilebasic.png" id="img_id3147576"><alt
id="alt_id3147576">Icon</alt></image>

We have tons of strings like this where the only translatable content
is "Icon". Luckily the translation for my language is quite similar so
with a keystroke and two mouse clicks I can quickly translate those,
but having to do it more than a hundred times is quite annoying since
all these could be just simplified to something like:

<placeholder>Icon</placeholder>

Or even better just:

Icon

Even if it is a massive effort and having in mind that it will require
retranslating thousands of strings I fully support simplifying the
strings using placeholders and removing any styling from the string if
possible. This should:

- reduce the number of strings,
- shorten the strings thus,
  - making it easier to translate,
  - making it faster to translate,
- reduce the possibility of typing mistakes,
- allow to increase use of translation memory.

IMHO these are big upsides making it worth the change.

Regards

#11

Þann mið 30.ágú 2017 09:14, skrifaði Martin Srebotnjak:

So, these changes will make 100+ l10n teams to re-translate/check those
touched strings, yes?

Well, these have been brewing for some time; think we have to live with
that. But of course we should make noise if there is a bunch of such
'cosmetic' modifications, all at once.
Meanwhile, we could think seriously about why the help/docs markup is so
complex (instead of being more CSS-like?) and why there's so little
separation of content and formatting - and what can be done???

I would also add why strings have so many tags when the only
translatable content is outside of the tags or in a single place in
the string, like:

<image src="cmd/sc_compilebasic.png" id="img_id3147576"><alt
id="alt_id3147576">Icon</alt></image>

We have tons of strings like this where the only translatable content
is "Icon". Luckily the translation for my language is quite similar so
with a keystroke and two mouse clicks I can quickly translate those,
but having to do it more than a hundred times is quite annoying since
all these could be just simplified to something like:

<placeholder>Icon</placeholder>

Or even better just:

Icon

Even if it is a massive effort and having in mind that it will require
retranslating thousands of strings I fully support simplifying the
strings using placeholders and removing any styling from the string if
possible. This should:

- reduce the number of strings,
- shorten the strings thus,
   - making it easier to translate,
   - making it faster to translate,
- reduce the possibility of typing mistakes,
- allow to increase use of translation memory.

IMHO these are big upsides making it worth the change.

Sorry for crashing the party but

     - documentation should be translatable article by article, not sentence by sentence. It's hard to distinguish title of a topic (article) from a regular text from a menu entry. It's not impossible but is not easy enough.

     - [know it's a black magic but] think we could all benefit from a system that could extract menu entries from UI in a manner of multidimensional arrays, apply some kind of ID for every (part/level) of them and than (by hand?) edit Help markup to cross-reference ID's with new pair of tags (XML can handle this) to markup such entries in Help files by levels (as such entries sometimes are separated with regular text). This way Help entries could be automatically replaced with corresponding entries form UI -- consistently!

Menu entries should be translated automatically, not two times by hand (UI and Help) regardless of translation memory. Now small change in UI results in massive change in Help. I don't don't have every UI menu entry in my head and hate to look up when translating Help (that's the reason I'm not doing it often). Feels like a waste of time. Inconsistency in translation is what really bothers me.

I know this would be massive change but why not at least try some brainstorming and see what gives...

I would like to see some comments from translators as in if this would be useful and from developers watching this list as in if this is even possible.

Thanks,

Kruno

#12

Hello Leandro, Hello everybody

Some comments on this thread:

Þann mið 30.ágú 2017 09:14, skrifaði Martin Srebotnjak:

So, these changes will make 100+ l10n teams to re-translate/check those
touched strings, yes?

Well, these have been brewing for some time; think we have to live with
that. But of course we should make noise if there is a bunch of such
'cosmetic' modifications, all at once.
Meanwhile, we could think seriously about why the help/docs markup is so
complex (instead of being more CSS-like?) and why there's so little
separation of content and formatting - and what can be done???

Given the complexity of the XML, it may be that the original developers
thought on using the XML contents in other kind of usage than simple
display. Other times, other visions. The reality is that this XML is
actually used only to display information.

My last blog post (*) was challenging the continuation of this XML.
AFAIK, except for one or 2 tags, HTML5 (or markdown) can replace this
XML and open the LibreOffice help content edition and update to mankind.

However, migrating XML to HTML (or markdown) is the easiest part of the
job. Dealing with the translation process is the hard part because it
involves the update of the tools that builds the po files as well as a
possible (perhaps not probable) stress on the translators to deal with
the change.

I would also add why strings have so many tags when the only
translatable content is outside of the tags or in a single place in
the string, like:

<image src="cmd/sc_compilebasic.png" id="img_id3147576"><alt
id="alt_id3147576">Icon</alt></image>

We have tons of strings like this where the only translatable content
is "Icon". Luckily the translation for my language is quite similar so
with a keystroke and two mouse clicks I can quickly translate those,
but having to do it more than a hundred times is quite annoying since
all these could be just simplified to something like:

<placeholder>Icon</placeholder>

Or even better just:

Icon

That particular case is one of the mistakes that never get fixed. The
word Icon is inside an <alt> tag, and should describe the meaning of the
icon when the image is not loaded (and this is part of the accessibility
specs). Someone got lazy when the image was inserted in the help file
long time ago....

Even if it is a massive effort and having in mind that it will require
retranslating thousands of strings I fully support simplifying the
strings using placeholders and removing any styling from the string if
possible. This should:

- reduce the number of strings,
- shorten the strings thus,
  - making it easier to translate,
  - making it faster to translate,
- reduce the possibility of typing mistakes,
- allow to increase use of translation memory.

IMHO these are big upsides making it worth the change.

For me, the worst case is the <switchinline><caseinlne> @#$%$#@ to
separate MAC from WIN and UNIX. It can be replaced by proper wording in
the sentence.

But re-translation is out of question. Current translations are an asset
of the L10N community and the LibreOffice software. Massive changes must
go invisible.

Besides, I also noticed the new Help pages and UI changes adds thousands
of words or reviews each 6 months, so a delicate balance between
evolution and translation is key for our L10N community, otherwise we
begin to see gaps in the translations (and thus loss of this "asset").

That is not to discard changes in the current markup. My last blog post
was an invitation to smoothly move away from XML to a more accessible
editing markup such as HTML5. By opening help edition to more
individuals, we can vastly improve its contents and quality.

Whatever is the solution to address our current issues, it has to be
smooth and bear the L10N community constraints.

As a side note, the XML is fully described in (**). It may help one to
understand the tags we see in Pootle.

Kind regards
Olivier

(*)
https://blog.documentfoundation.org/blog/2017/07/19/taming-libreoffice-help-system/

(**)
https://wiki.documentfoundation.org/Documentation/Understanding,_Authoring_and_Editing_Openoffice.org_Help/3