Re: Utility functions from gnome-libs
- From: Darin Adler <darin eazel com>
- To: Owen Taylor <otaylor redhat com>, <gtk-devel-list redhat com>
- Subject: Re: Utility functions from gnome-libs
- Date: Wed, 05 Jul 2000 11:40:26 -0700
On 3 Jul 2000, Owen Taylor wrote:
>>> enum {
>>> G_FILE_TEST_ISFILE=1<<0,
>>> G_FILE_TEST_ISLINK=1<<1,
>>> G_FILE_TEST_ISDIR=1<<2,
>>> G_FILE_TEST_EXISTS=(1<<0)|(1<<1)|(1<<2) /*any type of file*/
>>> };
>>>
>>> gboolean g_file_test (const char *filename, int test);
>>> gboolean g_file_exists (const char *filename);
>>>
>>> (These might alternatively be called g_file_race() ;-), but
>>> they are useful for a lot of things, and a note in the
>>> docs about security implications should be OK.)
Tim Janik <timj@gtk.org> writes:
>> no, that's not a good idea. instead, it'll probably not hurt to
>> simply implement g_file_test() in terms of:
>>
>> typedef enum
>> {
>> G_FILE_IS_FILE = 1 << 0,
>> G_FILE_IS_LINK = 1 << 1,
>> G_FILE_IS_DIR = 1 << 2,
>> G_FILE_EXISTS = G_FILE_IS_FILE | G_FILE_IS_LINK | G_FILE_IS_DIR
>> } GFileType;
>>
>> GFileType g_file_test (const char *filename);
on 7/5/00 11:27 AM, Owen Taylor at otaylor@redhat.com wrote:
> Yes, the enumeration should be named, not anonymous. Was there
> another objection or was that all?
I believe Tim's point was that all the tests are done atomically instead of
passing in the test you want to do, and this would somehow minimize the race
conditions.
-- Darin
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]