On Tue, 11 Feb 2003 23:01:48 +0100, Branko � ibej wrote:
>Joe Orton wrote:
>>--- filestat.c 7 Jan 2003 00:52:53 -0000 1.64
>>+++ filestat.c 11 Feb 2003 18:19:12 -0000
>>@@ -135,6 +135,8 @@
>> finfo->pool = thefile->pool;
>> finfo->fname = thefile->fname;
>> fill_out_finfo(finfo, &info, wanted);
>>+ if (thefile->buffered)
>>+ finfo->size += thefile->bufpos;
>> return (wanted & ~finfo->valid) ? APR_INCOMPLETE : APR_SUCCESS;
>> }
>> else {
>>
>>
>
>Hmm, does this patch actually work? I suspect the calculation would be
>slightly more involved -- after all, the buffer might not extend past
>the hard end-of-file at all.
And the buffer might contain read data, not written. This is the patch I'm
considering for the OS/2 code, it passes the File Info test:
Index: filestat.c
===================================================================
RCS file: /home/cvs/apr/file_io/os2/filestat.c,v
retrieving revision 1.34
diff -u -r1.34 filestat.c
--- filestat.c 7 Jan 2003 00:52:51 -0000 1.34
+++ filestat.c 13 Feb 2003 10:20:08 -0000
@@ -141,6 +141,12 @@
if (finfo->filetype == APR_REG) {
if (thefile->isopen) {
+ if (thefile->buffered && thefile->direction == 1) {
+ if (thefile->filePtr + thefile->bufpos > finfo->size) {
+ finfo->size = thefile->filePtr + thefile->bufpos;
+ }
+ }
+
return handle_type(&finfo->filetype, thefile->filedes);
}
} else {
--
______________________________________________________________________________
| Brian Havard | "He is not the messiah! |
| [EMAIL PROTECTED] | He's a very naughty boy!" - Life of Brian |
------------------------------------------------------------------------------
