On Thu, 28 Nov 2024, Pali Rohár wrote:
On Friday 29 November 2024 00:18:38 Martin Storsjö wrote:
On Tue, 19 Nov 2024, Pali Rohár wrote:
Currently mingw_vfwscanf.c files provides 3 functions:
mingw_swformat(), mingw_vfwscanf() and mingw_vswscanf().
Split this file into 3 files, one for each function.
Diff for this change from git is more human readable when showed by
git options -B and -D.
---
mingw-w64-crt/Makefile.am | 4 +-
.../{mingw_vfwscanf.c => mingw_swformat.c} | 46 +-
mingw-w64-crt/stdio/mingw_swformat.h | 63 +
mingw-w64-crt/stdio/mingw_vfwscanf.c | 1645 +----------------
mingw-w64-crt/stdio/mingw_vswscanf.c | 15 +
5 files changed, 102 insertions(+), 1671 deletions(-)
copy mingw-w64-crt/stdio/{mingw_vfwscanf.c => mingw_swformat.c} (97%)
create mode 100644 mingw-w64-crt/stdio/mingw_swformat.h
rewrite mingw-w64-crt/stdio/mingw_vfwscanf.c (99%)
create mode 100644 mingw-w64-crt/stdio/mingw_vswscanf.c
index 81cb46b661e9..5f4b2c08f395 100644
--- a/mingw-w64-crt/stdio/mingw_vfwscanf.c
+++ b/mingw-w64-crt/stdio/mingw_swformat.c
@@ -57,6 +57,8 @@
#include <locale.h>
#include <errno.h>
+#include "mingw_swformat.h"
+
#ifndef CP_UTF8
#define CP_UTF8 65001
#endif
@@ -80,19 +82,6 @@
#define IS_ALLOC_USED (USE_GNU_ALLOC | USE_POSIX_ALLOC)
-/* internal stream structure with back-buffer. */
-typedef struct _IFP
-{
- __extension__ union {
- void *fp;
- const wchar_t *str;
- };
- int bch[1024];
- unsigned int is_string : 1;
- int back_top;
- unsigned int seen_eof : 1;
-} _IFP;
-
static void *
get_va_nth (va_list argp, unsigned int n)
{
@@ -123,7 +112,7 @@ optimize_alloc (char **p, char *end, size_t alloc_sz)
}
static void
-back_ch (int c, _IFP *s, size_t *rin, int not_eof)
+back_ch (int c, _IFPW *s, size_t *rin, int not_eof)
{
if (!not_eof && c == WEOF)
return;
So here we're renaming the structure _IFP into _IFPW - is this a required
step? I guess this may have been the case if you initially considered using
one shared header for both narrow and wide versions of it, but as this
currently has two separate headers, I guess the struct renaming strictly
isn't necessary?
// Martin
Yea, maybe the renaming is not necessary. But I think it is better if we
do not have two different structures with same names in header files.
Yes, that's probably reasonable. Anyway, I think this aspect is fine, but
it would be good to mention it in a commit message (the fact that it's not
necessary but it is done to disambiguate the two variants of the struct).
If I don't have other issues with this patchset, I can amend that message
before pushing.
// Martin
_______________________________________________
Mingw-w64-public mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
