More checks in compare-locales, spot bad reports?
Axel Hecht
I've added some checks to compare-locales for .properties (bugs on .dtd
are already filed), and I was hoping you guys would chime in on the data
I got.
I ran this on a rather recent copy of 1.9.2, against firefox.
WARNING are warning messages, i.e., things worth looking at, but AFAICT
not technically errors.
ERROR are things that may very well just crash.
I'd be interested in warnings that should be errors, or errors that
should really just be warnings, or feedback in general.
Thanks, and fear a dashboard near you in space and time :-)
Axel
wladow
sk/toolkit/chrome/mozapps/extensions/extensions.properties
ERROR: argument 3 `S` obsolete at line 99, column 26 within
newAddonsNotificationMsg2
If I'm reading the error msg correctly, the third %S in the string is
considered obsolete? If so, it isn't obsolete for sure, the
newAddonsNotificationMsg2 is a pluralForm and we need a third form here.
Not sure why this is the only pluralForm string the cl is complaining about.
--
wladow
Axel Hecht
> There's just one error for sk:
> sk/toolkit/chrome/mozapps/extensions/extensions.properties
> ERROR: argument 3 `S` obsolete at line 99, column 26 within
> newAddonsNotificationMsg2
>
> If I'm reading the error msg correctly, the third %S in the string is
> considered obsolete? If so, it isn't obsolete for sure, the
> newAddonsNotificationMsg2 is a pluralForm and we need a third form here.
>
> Not sure why this is the only pluralForm string the cl is complaining
> about.
Good catch, gotta think about how I'll fix that. Thanks.
Axel
Mad Maks
> nl/dom/chrome
> appstrings.properties
> ERROR: argument 1 `S` obsolete at line 45, column 20 within confirmRepostPrompt
> layout/css.properties
> ERROR: Double ordered argument 1 at line 101, column 86 within PEPseudoClassNoClose
> nl:
> keys: 940
> unchanged: 615
> changed: 4580
> errors: 2
> 88% of entries changed
thanks, i have fixed them. But are they errors or warnings, these
"errors" where there for a long time (change in en-us at 5 jan 2008)
without breaking the builds
greetings
MM
Axel Hecht
I did some local tests, and I got "(null)" inserted for a missing
argument to S. It's not crashing, which is nice, but it's still an error.
Axel
smo
...
>
> I'd be interested in warnings that should be errors, or errors that
> should really just be warnings, or feedback in general.
>
> Thanks, and fear a dashboard near you in space and time :-)
>
> Axel
>
> 66 1KPrikažiPrenesi
Here's my check of warnings in sl material
browser/chrome/browser
feeds/subscribe.properties
1 WARNING: trailing argument 2 `S` missing at line 2, column 11
within addHandler
addHandler=Add "%S" (%S) as a Feed Reader?
tabbrowser.properties
2 WARNING: trailing argument 1 `S` missing at line 7, column 30
within tabs.closeWarningMultipleTabs
source: tabs.closeWarningMultipleTabs=You are about to close %S tabs.
Are you sure you want to continue?
dom/chrome/layout/xmlparser.properties
3 ERROR: argument 3 `d` should be `u`, argument 4 `d` should be
`u` at line 76, column 18 within XMLParsingError
source: XMLParsingError = XML Parsing Error: %1$S\nLocation:
%2$S\nLine Number %3$u, Column %4$u:
(translation already corrected before Axel's check, but not
committed)
toolkit/chrome/mozapps/profile/profileSelection.properties
4 WARNING: trailing argument 1 `S` missing at line 12, column 13
within pleaseSelect
source: pleaseSelect=Please select a profile to begin %S, or
create a new profile.
WARNING: trailing argument 3 `S` missing at line 15, column 15
within profileLocked2
source: profileLocked2=%S cannot use the profile "%S" because it
is in use.\n\nTo continue, close the running instance of %S or choose
a different profile.
Re three Warnings 1,2 and 4: the translation is correct, it's the
source that's wrong. Would it not make sense to correct the source and
propagate it through all the "wrong" localizations? Surely it would
not hurt to fix the source. Except if by some divine intervention en-
US versions do not crash on this kind of noneties.
Or, am I missing something?
PS: my en-US source is mozilla-1-9-2-3a6d01ad4e52
PPS: Let the dashboard be with you, Axel;)
Axel Hecht
Mind explaining why the source is wrong? Like,
http://mxr.mozilla.org/mozilla1.9.2/source/browser/components/feeds/src/WebContentConverter.js#589
is what I find to be feeding the content there, and it's very much
feeding two arguments.
Axel
Robert Kaiser
> de
> [...]
> toolkit/chrome/mozapps/update/updates.properties
> WARNING: trailing argument 1 `S` missing at line 24, column 18 within verificationError
That one is tricky.
en-US has:
verificationError=%S could not confirm the integrity of the update package.
de has:
verificationError=Die Integrität des Update-Pakets konnte nicht
sichergestellt werden.
Actually using the %S in the German string doesn't make too much sense,
it doesn't improve the meaning at all.
Robert Kaiser
Axel Hecht
That's the use case why I don't error on these but warn. I don't expect
the warnings to go to zero, with errors on the other hand I will.
Axel
Axel Hecht
trying again, if you get this twice, sorry. I guess I need to get a bit
more wordy to be allowed to attach the output of another pass.
I did a rather hacky approach to detect plurals, unless someone can come
up with something better, I scan the leading comment for
'Localization_and_Plurals' which is the catch phrase of the MDC link
that the plural localization notes all seem to have.
Another improvement I added locally is that l10n-merge actually strips
broken entities now, so the builds should work bustage free, at least as
far as the tests can figure that out.
More comments on the .properties output are welcome
Axel
Simon Paquet
> trying again, if you get this twice, sorry. I guess I need to get a bit
> more wordy to be allowed to attach the output of another pass.
Axel,
attaching files here is not really helpful, at least not to me and to
pretty much everyone else, who reads this group via NNTP, because
all attachments get filtered out there in non-binary groups.
Can't you just upload those attachments to a place somewhere on
l10n.mozilla.org?
Cya
Simon
--
Thunderbird/Calendar Localisation (L10n) Coordinator
Thunderbird l10n blog: http://thunderbird-l10n.blogspot.com
Calendar website maintainer: http://www.mozilla.org/projects/calendar
Calendar developer blog: http://weblogs.mozillazine.org/calendar
Mark Banner
> Axel Hecht wrote on 25. Mar 2010:
>
>> trying again, if you get this twice, sorry. I guess I need to get a
>> bit more wordy to be allowed to attach the output of another pass.
>
> Axel,
>
> attaching files here is not really helpful, at least not to me and to
> pretty much everyone else, who reads this group via NNTP, because
> all attachments get filtered out there in non-binary groups.
Weird, I'm reading Axel's messages on NNTP and I'm seeing the attachments.
Mark.
smo
> is what I find to be feeding the content there, and it's very much
> feeding two arguments.
>
> Axel
What are you feeding the programs with?!
Check my reference: There's not one single %Sn argument (strictly %S
arguments only) in the en-US files your analysis has tagged as faulty
in the sl localization. The reference is the official 87.1MB of
Mozilla stuff - I wanted to exclude any chance of getting the wrong
version, so I pulled the head version. If this was wrong, i.e. if we
refer to different en-US sources, then PLEASE add a reference, mother-
of-them-all en-US branch to l10n releases, so that there's no more
misunderstandings about it.
Regards
smo
Axel Hecht
http://mxr.mozilla.org/mozilla1.9.2/source/browser/locales/en-US/chrome/browser/feeds/subscribe.properties?mark=2-2#2
highlighting the line in question.
Axel