Re: [systemd-devel] [RFC PATCH 2/2] localed: add LANGUAGE= fallback when LANG= is specified
On Mon, Feb 02, 2015 at 03:10:37PM +0100, Zbigniew Jędrzejewski-Szmek wrote: On Mon, Feb 02, 2015 at 11:56:39AM +0100, Martin Pitt wrote: Zbigniew Jędrzejewski-Szmek [2015-01-31 20:20 +0100]: I think the implementation is fine, since it is rather trivial, but I'm less certain about the implications of setting LANGUAGE in addtion to LANG. That's totally fine -- it's precisely what $LANGUAGE is meant for. One must be aware that it is a GNU extension, i. e. software like thunderbird won't respect it and thus just use $LANG (in which case you are no worse than now). But pretty much all native Linux sofware does and thus will behave as intended. We've used this schema in Ubuntu since around 2005 [1]. Thanks. Incomplete support sounds fine. Now, I wish such a fallback list would be in upstream glibc so that OSes, systemd etc. wouldn't have to repeat and maintain it; but while [1] looks quite complex, it hasn't really changed in years. Looking at the list, I'm wondering about using en_GB as the fallback for most locales. I guess that for European countries it makes some sense, but for Japanese, Kazakh, and many other places, I'd expect en_US to be better last choice. Especially that en_US translation coverage is more complete than en_GB. Also, I'd expect a fallback to Russian before English for Kazakh, it's their second official language. If this patch goes in, I'm happy to augment the list according to [1] and then move our language-selector to use that instead, to reduce duplication. Great. I pushed the patches now. It would be great if you could add more mappings. Zbyszek ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
Re: [systemd-devel] [RFC PATCH 2/2] localed: add LANGUAGE= fallback?when LANG= is specified
On Mon, Feb 02, 2015 at 07:11:04AM +, Jan Janssen wrote: Zbigniew Jędrzejewski-Szmek zbyszek at in.waw.pl writes: I think the implementation is fine, since it is rather trivial, but I'm less certain about the implications of setting LANGUAGE in addtion to LANG. Zbyszek Isn't this something that should be in glibc's lang/locale handling code? They probably could in principle, but those settings are somewhat arbitrary and at a higher level than libc usually operates. We are already taking care of configuration of a lot of related stuff, so this seems a natural extension. Zbyszek ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
Re: [systemd-devel] [RFC PATCH 2/2] localed: add LANGUAGE= fallback when LANG= is specified
On Mon, Feb 02, 2015 at 11:56:39AM +0100, Martin Pitt wrote: Zbigniew Jędrzejewski-Szmek [2015-01-31 20:20 +0100]: I think the implementation is fine, since it is rather trivial, but I'm less certain about the implications of setting LANGUAGE in addtion to LANG. That's totally fine -- it's precisely what $LANGUAGE is meant for. One must be aware that it is a GNU extension, i. e. software like thunderbird won't respect it and thus just use $LANG (in which case you are no worse than now). But pretty much all native Linux sofware does and thus will behave as intended. We've used this schema in Ubuntu since around 2005 [1]. Thanks. Incomplete support sounds fine. Now, I wish such a fallback list would be in upstream glibc so that OSes, systemd etc. wouldn't have to repeat and maintain it; but while [1] looks quite complex, it hasn't really changed in years. Looking at the list, I'm wondering about using en_GB as the fallback for most locales. I guess that for European countries it makes some sense, but for Japanese, Kazakh, and many other places, I'd expect en_US to be better last choice. Especially that en_US translation coverage is more complete than en_GB. Also, I'd expect a fallback to Russian before English for Kazakh, it's their second official language. If this patch goes in, I'm happy to augment the list according to [1] and then move our language-selector to use that instead, to reduce duplication. Great. Zbyszek ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
Re: [systemd-devel] [RFC PATCH 2/2] localed: add LANGUAGE= fallback when LANG= is specified
Zbigniew Jędrzejewski-Szmek [2015-01-31 20:20 +0100]: I think the implementation is fine, since it is rather trivial, but I'm less certain about the implications of setting LANGUAGE in addtion to LANG. That's totally fine -- it's precisely what $LANGUAGE is meant for. One must be aware that it is a GNU extension, i. e. software like thunderbird won't respect it and thus just use $LANG (in which case you are no worse than now). But pretty much all native Linux sofware does and thus will behave as intended. We've used this schema in Ubuntu since around 2005 [1]. Now, I wish such a fallback list would be in upstream glibc so that OSes, systemd etc. wouldn't have to repeat and maintain it; but while [1] looks quite complex, it hasn't really changed in years. If this patch goes in, I'm happy to augment the list according to [1] and then move our language-selector to use that instead, to reduce duplication. Martin [1] http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/vivid/language-selector/vivid/view/head:/data/languagelist -- Martin Pitt| http://www.piware.de Ubuntu Developer (www.ubuntu.com) | Debian Developer (www.debian.org) ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
Re: [systemd-devel] [RFC PATCH 2/2] localed: add LANGUAGE= fallback when LANG= is specified
On Sat, 31.01.15 20:20, Zbigniew Jędrzejewski-Szmek (zbys...@in.waw.pl) wrote: For the entries listed in the first column of language-fallback-map, the entry from the second column will be used for LANGUAGE=, if LANGUAGE= is not explicitly specified. Looks conceptually OK to me, sounds like a useful addition! Thanks! Lennart -- Lennart Poettering, Red Hat ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
Re: [systemd-devel] [RFC PATCH 2/2] localed: add LANGUAGE= fallback when LANG= is specified
Zbigniew Jędrzejewski-Szmek zbyszek at in.waw.pl writes: I think the implementation is fine, since it is rather trivial, but I'm less certain about the implications of setting LANGUAGE in addtion to LANG. Zbyszek Isn't this something that should be in glibc's lang/locale handling code? ___ systemd-devel mailing list systemd-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/systemd-devel
[systemd-devel] [RFC PATCH 2/2] localed: add LANGUAGE= fallback when LANG= is specified
For the entries listed in the first column of language-fallback-map,
the entry from the second column will be used for LANGUAGE=, if
LANGUAGE= is not explicitly specified.
https://bugzilla.redhat.com/show_bug.cgi?id=624158
---
This fixes a long-standing bug where users of a dialect locale would like
to fallback to a more general one for translations. For example, zh_HK to
zh_CN, etc.
I think the implementation is fine, since it is rather trivial, but I'm
less certain about the implications of setting LANGUAGE in addtion to
LANG.
Zbyszek
Makefile.am | 1 +
src/locale/localed.c | 76 +---
2 files changed, 68 insertions(+), 9 deletions(-)
diff --git a/Makefile.am b/Makefile.am
index e3ba11c8c0..f359da7154 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -190,6 +190,7 @@ AM_CPPFLAGS = \
-DSYSTEM_SHUTDOWN_PATH=\$(systemshutdowndir)\ \
-DSYSTEM_SLEEP_PATH=\$(systemsleepdir)\ \
-DSYSTEMD_KBD_MODEL_MAP=\$(pkgdatadir)/kbd-model-map\ \
+ -DSYSTEMD_LANGUAGE_FALLBACK_MAP=\$(pkgdatadir)/language-fallback-map\
\
-DX_SERVER=\$(bindir)/X\ \
-DUDEVLIBEXECDIR=\$(udevlibexecdir)\ \
-DPOLKIT_AGENT_BINARY_PATH=\$(bindir)/pkttyagent\ \
diff --git a/src/locale/localed.c b/src/locale/localed.c
index 529a9abfd6..3a11992f6a 100644
--- a/src/locale/localed.c
+++ b/src/locale/localed.c
@@ -512,7 +512,9 @@ static const char* strnulldash(const char *s) {
return isempty(s) || streq(s, -) ? NULL : s;
}
-static int read_next_mapping(FILE *f, unsigned *n, char ***a) {
+static int read_next_mapping(const char* filename,
+ unsigned min_fields, unsigned max_fields,
+ FILE *f, unsigned *n, char ***a) {
assert(f);
assert(n);
assert(a);
@@ -521,6 +523,7 @@ static int read_next_mapping(FILE *f, unsigned *n, char
***a) {
char line[LINE_MAX];
char *l, **b;
int r;
+size_t length;
errno = 0;
if (!fgets(line, sizeof(line), f)) {
@@ -541,8 +544,9 @@ static int read_next_mapping(FILE *f, unsigned *n, char
***a) {
if (r 0)
return r;
-if (strv_length(b) 5) {
-log_error(Invalid line SYSTEMD_KBD_MODEL_MAP:%u,
ignoring., *n);
+length = strv_length(b);
+if (length min_fields || length max_fields) {
+log_error(Invalid line %s:%u, ignoring., filename,
*n);
strv_free(b);
continue;
@@ -579,7 +583,7 @@ static int vconsole_convert_to_x11(Context *c, sd_bus *bus)
{
_cleanup_strv_free_ char **a = NULL;
int r;
-r = read_next_mapping(f, n, a);
+r = read_next_mapping(SYSTEMD_KBD_MODEL_MAP, 5,
UINT_MAX, f, n, a);
if (r 0)
return r;
if (r == 0)
@@ -677,7 +681,7 @@ static int find_legacy_keymap(Context *c, char
**new_keymap) {
_cleanup_strv_free_ char **a = NULL;
unsigned matching = 0;
-r = read_next_mapping(f, n, a);
+r = read_next_mapping(SYSTEMD_KBD_MODEL_MAP, 5, UINT_MAX, f,
n, a);
if (r 0)
return r;
if (r == 0)
@@ -752,6 +756,35 @@ static int find_legacy_keymap(Context *c, char
**new_keymap) {
return 0;
}
+static int find_language_fallback(const char *lang, char **language) {
+_cleanup_fclose_ FILE *f = NULL;
+unsigned n = 0;
+
+assert(language);
+
+f = fopen(SYSTEMD_LANGUAGE_FALLBACK_MAP, re);
+if (!f)
+return -errno;
+
+for (;;) {
+_cleanup_strv_free_ char **a = NULL;
+int r;
+
+r = read_next_mapping(SYSTEMD_LANGUAGE_FALLBACK_MAP, 2, 2, f,
n, a);
+if (r = 0)
+return r;
+
+if (streq(lang, a[0])) {
+assert(strv_length(a) == 2);
+*language = a[1];
+a[1] = NULL;
+return 1;
+}
+}
+
+assert_not_reached(should not be here);
+}
+
static int x11_convert_to_vconsole(Context *c, sd_bus *bus) {
bool modified = false;
int r;
@@ -841,9 +874,10 @@ static int method_set_locale(sd_bus *bus, sd_bus_message
*m, void *userdata, sd_
Context *c = userdata;
_cleanup_strv_free_ char **l = NULL;
char **i;
+const char *lang = NULL;
int interactive;
bool modified = false;
-bool passed[_LOCALE_MAX] = {};
+bool have[_LOCALE_MAX] = {};
int p;
int r;
@@ -867,7 +901,10
