Pietro Cerutti [Tue, 29 May 2018 15:39:23 +0000 (15:39 +0000)]
Avoid regenerating hcache/hcversion.h after building objects in hcache
Specifying the hcache directory asqa dependency for hcache/hcversion.h
means that the latter will be regenerated whenever the hcache directory
is changed, such as when object files are written into it. The net
effect is that "make; make" regenerates hcache/hcversion.h twice.
This commit fixes this problem by explicitely creating the hcache
directory as part of the hcache/hcversion.h target.
Pietro Cerutti [Tue, 29 May 2018 13:29:02 +0000 (14:29 +0100)]
Add the mutt_ch_check API (#1139)
* Add the mutt_ch_check API
This adds the mutt_ch_check API that can be used to check whether a
string can be converted between character sets. It is stricter than
mutt_ch_convert_string in that it doesn't apply any replacement chars to
unconvertible characters and doesn't return the converted string.
And i want to make clear for that:if i place %d into msgstr[0],then
msgid "Print tagged attachment" how to translate?Under my understand,if
program invokes msgid, it will display "打印 件已标记的附件",just left a
space.
Reis Radomil [Tue, 22 May 2018 14:18:53 +0000 (14:18 +0000)]
Update po files to match new code base
Most changes are splitting up plural constructions into more "saner"
versions, e.g. _("message(s)") becomes ngettext("message", "messages",
n). In all the cases we used the old translation for the new pluralised
translation (in all plural forms). This might not be correct but yields
the same result as before the change to the end user.
Reis Radomil [Tue, 22 May 2018 14:16:12 +0000 (14:16 +0000)]
opcodes.h: Fix plural messages with "<word>(s)"
Due to the implementation details we do not know whether we work with
zero, 1, 12, ... messages/files/.... So in English we use normal plural.
We leave it up to the translator to choose a fitting translation. Some
language might have other means to express this situation.
Reis Radomil [Tue, 22 May 2018 14:16:12 +0000 (14:16 +0000)]
curs_main.c: Fix plural messages with "messages"
Due to the implementation details we do not know whether we
delete/undelete/mark/... zero, 1, 12, ... messages. So in English we use
"messages". We leave it up to the translator to choose a fitting
translation. Some language might have other means to express this
situation.
Reis Radomil [Tue, 22 May 2018 14:16:12 +0000 (14:16 +0000)]
pager.c: Fix plural messages with "messages"
Due to the implementation details we do not know whether we
delete/undelete zero, 1, 12, ... messages. So in English we use
"messages". We leave it up to the translator to choose a fitting
translation. Some language might have other means to express this
situation.
Reis Radomil [Tue, 22 May 2018 14:16:12 +0000 (14:16 +0000)]
handler.c: Partly fix plural message about attachment size
Use ngettext() to choose a (more) correct plural form for 'byte'
depending on the attachment size. There is a caveat though: We do not
print the size as number but as a prettified number, e.g. 2048 will be
printed as 2K. Some languages might inflect 'byte' different depending
on whether it is '1K byte' or '1024'. Sadly there is nothing we can do
about that in the moment (besides not using the prettified version but
that is not an option).
Due to the implementation details (e.g. some passwords are managed by
gpg-agent) we cannot know whether we forgot zero, 1, 12, ... passwords.
So in English we use a generic plural. Different languages might have
other means, so we leave the decision what to choose to the translator.
to populate the buffy list when *first* opening a connection to foo.com,
but not on subsequent back-and-forth between IMAP servers.
The new code moves the LSUB call - and thus the population of the buffy
list - to just before a mailbox is selected.
This fixes the above-mentioned issue, at the cost of additional LSUB
calls when switching between mailboxes on the same server. Given that
the list of subscribed mailboxes is expected to be in the tens, I think
this is a reasonable tradeoff.
Marius Gedminas [Tue, 15 May 2018 12:03:16 +0000 (15:03 +0300)]
Update lt_LT translation
Mostly reviewing fuzzy entries in the first 60% of the file, but I also
added some missing translations and reviewed some older translations for
consistency.
* mutt/file.c: allow mutt_file_mkstemp to return a filename
modify mutt_file_mkstemp so that a caller may pass in a character
buffer that will be filled with the temporary filename. When a
non-null buffer is provided, mutt_file_mkstemp will *not* unlink the
temporary file and will leave cleanup to the caller.
ci_send_message: copy the x_original_to address list instead of using it
If an end user cancels the edit_envelope action, then cleanup is called.
When using directly the x_original_to address list directly, this
cleanup stage did actually clean the x_original_to pointer. Should the
user reply again to the same message, then the pointer would point to 0
and then any "unsafe" operation on it would trigger a SIGSEGV. Copying
it solves this issue safely.
Tobias Angele [Sat, 12 May 2018 10:12:42 +0000 (12:12 +0200)]
github: remove double checking question
`Are there points in the code the reviewer needs to double check?` is mostly providing no benefits. When a PR author wants the reviewer to double check sth in particular, they can somewhere.
Additionally, passing tests should (obviously) also be part of the acceptance criteria.