OpenStack Compute (nova)

Code review comment for lp:~citrix-openstack/nova/xenapi-glance-2

xenapi-glance-2
Merge into trunk

Revision history for this message

Armando Migliaccio (armando-migliaccio) wrote on 2011-01-14:

Thanks, that explains it.

I agree with you that we need to fix this, but I personally don't see it as much of an issue as most of the log strings are made of nouns, rarely with verbs and adjectives. This makes the translation pretty straightforward.

Since it affects every log in trunk I also think it would be better to have a trunk overhaul in a separate branch.

By the way, some of the languages (i.e. Japanese) have already been translated 100%...does this mean that those translations have to be reviewed?

> > can you elaborate a bit more? I see lots of localized strings with multiple
> > formatting placeholders and no use of dictionaries.
>
> Yes, and these will have to be corrected at some point.
>
> The problem is that translations are not word-for-word; a phrase in different
> languages may order the words differently. Probably the most common example
> would be adjectives coming before the noun in English ("the white house"), but
> after the noun in others (e.g., Spanish: "la casa blanca"). If your code looks
> like:
>
> color = "white"
> thing = "house"
> print _("The %s %s") % (color, thing)
>
> ... it will print "The white house" in English, but the Spanish will print "La
> blanca casa", which is wrong. You need to use a mapping for the formatting, so
> that the code above would read:
>
> color = "white"
> thing = "house"
> print _("The %(color)s %(thing)s") % locals()
>
> Yeah, I know that this is a weak example, since the color and thing are still
> in English, but it's just designed to demonstrate how positional substitution
> is to be avoided in localization strings.
>
> If you run xgettext on a file that uses tuples for multiple substitutions, it
> will emit the following:
> "warning: 'msgid' format string with unnamed arguments cannot be properly
> localized: The translator cannot reorder the arguments. Please consider using
> a format string with named arguments, and a mapping instead of a tuple for the
> arguments."
>
> I hope that that helps.

Thanks, that explains it.

Since it affects every log in trunk I also think it would be better to have a trunk overhaul in a separate branch.

By the way, some of the languages (i.e. Japanese) have already been translated 100%...does this mean that those translations have to be reviewed?

> > can you elaborate a bit more? I see lots of localized strings with multiple
> > formatting placeholders and no use of dictionaries.
> 
> Yes, and these will have to be corrected at some point.
> 
> The problem is that translations are not word-for-word; a phrase in different
> languages may order the words differently. Probably the most common example
> would be adjectives coming before the noun in English ("the white house"), but
> after the noun in others (e.g., Spanish: "la casa blanca"). If your code looks
> like:
> 
> color = "white"
> thing = "house"
> print _("The %s %s") % (color, thing)
> 
> ... it will print "The white house" in English, but the Spanish will print "La
> blanca casa", which is wrong. You need to use a mapping for the formatting, so
> that the code above would read:
> 
> color = "white"
> thing = "house"
> print _("The %(color)s %(thing)s") % locals()
> 
> Yeah, I know that this is a weak example, since the color and thing are still
> in English, but it's just designed to demonstrate how positional substitution
> is to be avoided in localization strings.
> 
> If you run xgettext on a file that uses tuples for multiple substitutions, it
> will emit the following:
> "warning: 'msgid' format string with unnamed arguments cannot be properly
> localized: The translator cannot reorder the arguments. Please consider using
> a format string with named arguments, and a mapping instead of a tuple for the
> arguments."
> 
> I hope that that helps.

« Back to merge proposal