Verbatim upload pares down the contents of the uploaded files
smo
PoEdit and some other tools - for instance I have no time and temper
to go through 30+ segments with extra blanks after the final period, I
use Notepad++ for that.
I lokalize into Slovenian, which needs nplurals=4 form, as for
instance in the following example:
#: controllers/components/amo.php:714
#, php-format
msgid "%1$s %2$s file"
msgid_plural "%1$s %2$s files"
msgstr[0] "%1$s %2$s datoteka"
msgstr[1] "%1$s %2$s datoteki"
msgstr[2] "%1$s %2$s datoteke"
msgstr[3] "%1$s %2$s datotek"
well, it looks to me (it took me several tries to figure it out, and
it is quite possible, this not the end of the story) that the upload
decides for nplurals=2 rule and "adjusts" everything like this:
#: controllers/components/amo.php:714
#, php-format
msgid "%1$s %2$s file"
msgid_plural "%1$s %2$s files"
msgstr[0] "%1$s %2$s datoteka"
msgstr[1] "%1$s %2$s datotek"
in other words, after uploading the material, I am back at square one,
as regards nplurals errors.
Can somebody please tell me the SVN address for the repository, so
that I can circumnavigate Verbatim?
regards
Vito
Julen
>
> Can somebody please tell me the SVN address for the repository, so
> that I can circumnavigate Verbatim?
That is:
http://svn.mozilla.org/addons/trunk/site/app/locale/
and the viewvc interface is available at:
http://viewvc.svn.mozilla.org/vc/addons/trunk/site/app/locale/
But I can't see any 'sl' directory there... :/
Julen.
smo
False alarm!
When uploading, I did not ask for overwrite, so the contents were
merged - somehow / beyond recognition.
I'd rate myself as a normal (in Gaussian sense) user, with one right
hand, and one left hand and the usual/normal count of thumbs..
Whatever happened, it should not;\
... btw, I found the SVN myself and yes, Julien, no sl ... and one of
the files in my Verbatim repository is called z-messages (and not r-
messages as elsewhere).
One last point, regarding contents of these files:
Active Daily Users
%1$s active daily users
<em>%s</em> Active Daily Users
<li><em>%s</em> Active Daily Users</li>
{0} Active Daily Users
is it strictly necessary to burden us with this kind of BS variants!?
regards
Vito
flod
> Can somebody please tell me the SVN address for the repository, so
> that I can circumnavigate Verbatim?
>
Here's the repository (but I don't see "sl")
http://viewvc.svn.mozilla.org/vc/addons/trunk/site/app/locale/
Francesco
smo
> 2010/12/27 smo <vito.smo...@gmail.com>
thx francesco!
note that all html items (<p />,<li />, <h? /> are mangled, so watch
out!
wish you a little late Buon Natale and a happy New Year! Thanks for
your help in 2010.
Vito
smo
Oops, i was talking about http://www.frenchmozilla.fr/glossaire/TMX/memoire_en-US_{AB_CD}.tmx-
Appologies for the confusion
Besnik Bleta
L10n dashboard reports no error for my locale, while users had an error
from the nightly build,
complaining about a *&securityOverride.**warningContent;* related error.
Checking against original found that:
<!ENTITY securityOverride.warningContent "
<p>You should not add an exception if you are using an internet
connection that you do not trust completely or if you are not used to
seeing a warning for this server.</p>
<xul:button
xmlns:xul='http://www.mozilla.org/keymaster/gatekeeper/there.is.only.xul' id='getMeOutOfHereButton'
label='&securityOverride.getMeOutOfHereButton;'/>
<xul:button
xmlns:xul='http://www.mozilla.org/keymaster/gatekeeper/there.is.only.xul' id='exceptionDialogButton'
label='&securityOverride.exceptionButtonLabel;'/>
">
has become:
<!ENTITY securityOverride.warningContent "
<p>You should not add an exception if you are using an internet
connection that you do not trust completely or if you are not used to
seeing a warning for this server.</p>
<button
id='getMeOutOfHereButton'>&securityOverride.getMeOutOfHereButton;</button>
<button
id='exceptionDialogButton'>&securityOverride.exceptionButtonLabel;</button>
">
You might like to check your translation as well, unless I've missed a
previously stated bug here...
cheers
Dwayne Bailey
> On Dec 27, 4:36 pm, Julen<jul...@gmail.com> wrote:
>> al., 2010.eko aberen 27a 16:22(e)an, smo(e)k idatzi zuen:
>>
>>
>>
>>> Can somebody please tell me the SVN address for the repository, so
>>> that I can circumnavigate Verbatim?
>> That is:http://svn.mozilla.org/addons/trunk/site/app/locale/
>>
>> and the viewvc interface is available at:http://viewvc.svn.mozilla.org/vc/addons/trunk/site/app/locale/
>>
>> But I can't see any 'sl' directory there... :/
>>
>> Julen.
> False alarm!
>
> When uploading, I did not ask for overwrite, so the contents were
> merged - somehow / beyond recognition.
helps you understand how Verbatim works with plural info and merging
mode. And I leave one task for you to try track down the errant 2 form
to make sure this doesn't happen again.
By "beyond recognition" I assume you only mean the plural entries not
the rest of the merge.
Merging is the least destructive way of uploading and thus its the
default. Merge mode will do the following: for anything untranslated it
will use your translation, for anything with existing translations it
will add your translation as a suggestion.
Using overwrite was the correct approach in your case. You have
complete control of what Verbatim accepts and allows you to change
everything including the plural header.
But that doesn't address the plural issue. It happened for these
possible reasons:
1. The SL plural equation is incorrect in Verbatim. When first
translating in Verbatim it sets the plural header if it is blank.
This seems unlikely since in checking the plural info we have for
SL it correctly lists SL plurals as 4 form. But it wouldn't hurt
to check that setting for SL in Verbatim so that new translations
get the correct form.
2. The actual SL file was incorrect i.e. using 2 form, when first
uploaded. Verbatim won't use its knowledge or 4 forms, we assume
that someone knew what they where doing when they uploaded 2 forms
and we don't change it. Believe it or not there are languages
that still can't agree on the correct plural form, so we take the
most diplomatic position and don't fiddle with plural forms that
users have defined. If its set to 2 we won't change it to 4.
Thus when you upload in merge mode we won't overwrite the plural
information, so Verbatim would simply convert your 4 form to 2 form.
You solved the problem correctly by using overwrite in the upload. But
I would suggest checking to find out where that two form came from.
It's possible that it came from an incorrect setting for SL plural info
in Verbatim. But more likely it came from an incorrectly set plural
header sometime during the translation of that file. So if there are
other people on your team please check that they don't have 2 plurals
set in their tools.
Hope that helps.
--
regards
Dwayne
smo
> On 2010-12-27 17:57,smowrote:> >> al., 2010.eko aberen 27a 16:22(e)an,smo(e)k idatzi zuen:
Thanks, Dwayne - a little belatedly but still
My biggest fault was not forcing the overwrite - it comes from
learning by doing. Of course it make sense, NOT to make it a default.
As regards plurals: you're damned if you follow the rules, you're
damned if you don't. We Slovenians are VERY proud (g) of our dual and
three plural forms, so I have to put my money, where my mouth is. But
I am considering writing a script to propagate the extra msgstr
entries automatically.
Hint (or rather a question): the diagnostics is impressive. Can I
adjust the level? Untranslated segments are more serious than double
blanks, and screwed-up parameters (mouch) more serious than missing
periods. It reminds me somehow of that pilot, who (with so many lights
blinking in his cockpit) decided to ignore them all. Including the one
yelling "landing gear not extracted".
Regards
smo
smo
> On 2010-12-27 17:57,smowrote:
>
>
>
The initial disconnect between project SVN and verbatim contents
amounted to more than 8000 words, which is - in view of the total
28482 words - awesome, but awesome the wrong way. I can relax,
however, because these ~8341 to-be-translated words are not as big as
they seem:
Segments Words
Repetitions: 29 153
Exact match: 3639 19988
95%-100%: 756 5788
85%-94%: 13 249
75%-84%: 11 87
50%-74%: 127 1896
No match: 18 321
(fuzzy match statistics by OmegaT):
in other words ~70% (5788) are minimal changes and easy to identify
via fuzzy matching. The real work are the remaining 2553 words - a
just penalty for my share of screw-ups.
regards
smo
Regards
smo
Dwayne Bailey
> My biggest fault was not forcing the overwrite - it comes from
> learning by doing. Of course it make sense, NOT to make it a default.
>
> As regards plurals: you're damned if you follow the rules, you're
> damned if you don't. We Slovenians are VERY proud (g) of our dual and
> three plural forms, so I have to put my money, where my mouth is. But
> I am considering writing a script to propagate the extra msgstr
> entries automatically.
>
> Hint (or rather a question): the diagnostics is impressive. Can I
> adjust the level? Untranslated segments are more serious than double
> blanks, and screwed-up parameters (mouch) more serious than missing
> periods. It reminds me somehow of that pilot, who (with so many lights
> blinking in his cockpit) decided to ignore them all. Including the one
> yelling "landing gear not extracted".
I agree with you on this issue. A long time ago I classified the test
into Critical, Functional, Cosmetic and Extraction see
http://translate.sourceforge.net/wiki/toolkit/pofilter_tests#test_classification
"Critical" are the landing gear type errors and I check these for most
of the languages that we maintain and leave the others to the rest of
the translation team to validated.
We just haven't had time to get that into Pootle.
However, one important thing we did add with 2.1 is the ability to mark
false positives. This is important for the non-critical tests to remove
the noise. Try those if you want to reduce the issues. Also not that
we can turn of tests for any language. We need to do that in the code
but once done it replicates to all users of Pootle. We can also
customise the tests to adapt for specific SL punctuation, etc. Friedel
is the right person to speak to to get that into the code.
--
cheers
Dwayne
Dwayne Bailey
On 2011-01-16 22:20, Dwayne Bailey wrote:
>
>
> On 2011-01-16 09:34, smo wrote:
>> learning by doing. Of course it make sense, NOT to make it a default.
>>
>> As regards plurals: you're damned if you follow the rules, you're
>> damned if you don't. We Slovenians are VERY proud (g) of our dual and
>> three plural forms, so I have to put my money, where my mouth is. But
>> I am considering writing a script to propagate the extra msgstr
>> entries automatically.
>>
>> Hint (or rather a question): the diagnostics is impressive. Can I
>> adjust the level? Untranslated segments are more serious than double
>> blanks, and screwed-up parameters (mouch) more serious than missing
>> periods. It reminds me somehow of that pilot, who (with so many lights
>> blinking in his cockpit) decided to ignore them all. Including the one
>> yelling "landing gear not extracted".
>
> I agree with you on this issue. A long time ago I classified the test
> into Critical, Functional, Cosmetic and Extraction see
> http://translate.sourceforge.net/wiki/toolkit/pofilter_tests#test_classification
>
> "Critical" are the landing gear type errors and I check these for most
> of the languages that we maintain and leave the others to the rest of
> the translation team to validated.
>
> We just haven't had time to get that into Pootle.
>
> However, one important thing we did add with 2.1 is the ability to
> mark false positives. This is important for the non-critical tests to
> remove the noise. Try those if you want to reduce the issues. Also
> not that we can turn of tests for any language. We need to do that in
> the code but once done it replicates to all users of Pootle.
Also note that we can turn off tests for any language.