* Andrew Dunstan:
On 04/09/2016 08:43 AM, Christian Ullrich wrote:
Michael Paquier wrote:
I don't think that's good to use malloc in those code paths, and I
+#if (_MSC_VER >= 1900)
+ uint32 cp;
+ if (GetLocaleInfoEx(ctype,
+ LOCALE_IDEFAULTANSICODEPAGE | LOCALE_RETURN_NUMBER,
+ (LPWSTR) &cp, sizeof(cp) / sizeof(WCHAR)) > 0)
+ r = malloc(16); /* excess */
+ if (r != NULL)
+ sprintf(r, "CP%u", cp);
I don't think we need to be too precious about saving a byte or two
here. Can one or other of you please prepare a replacement patch based
in this discussion?
Sorry, I don't think I'm up to that (at least not for another week or
so). I have to read up on the issue first; right now I'm not sure what
exactly the problem is.
What I can say is that the existing patch needs more work, because
GetLocaleInfoEx() expects a locale name ("de-DE") as its first argument,
but the patched win32_langinfo() is giving it a locale identifier
("German_Germany.1252"). At least it does for me.
I have not gone through the code sufficiently closely to find out where
the argument to win32_langinfo() originates, but I will when I can.
As for missing code page information in the _locale_t type, ISTM it
isn't hidden any better in UCRT than it was before:
/* Set locale with nonstandard code page */
_locale_t loc = _create_locale(LC_ALL, "German_Germany.850");
__crt_locale_data_public* pub =
printf("CP: %d\n", pub->_locale_lc_codepage); // "CP: 850"
Sent via pgsql-hackers mailing list (email@example.com)
To make changes to your subscription: