Hi,
On 30-11-16 02:16, Antonio Quartulli wrote:
> Do I really need to split declaration and definition? Or can I just
> doxy-document the definition in the same place where it is now?
Well, either add a declaration at the top of the file (before the first
usage), or move the function up to before the first usage and document
it without splitting it.
>> This can be done without the 'reload' argument too, something like:
>>
>> /* if something goes wrong with stat(), we'll store 0 as mtime */
>> platform_stat_t crl_stat = {0};
>>
>> if (crl_file_inline)
>> {
>> /*
>> * an inline CRL can't change at runtime, therefore there is no
>> need to
>> * reload it. It will be reloaded upon config change + SIGHUP.
>> */
>> if (ssl_ctx->crl_last_mtime.tv_sec)
>> return;
>> else
>> crl_stat.st_mtime = now; /* Set dummy mtime */
>> }
>> else
>> {
>> platform_stat (crl_file, &crl_stat);
>> }
>>
>> /*
>> * If this is not a reload, update the CRL right away.
>> * Otherwise, update only if the CRL file was modified after the last
>> reload.
>> * Note: Windows does not support tv_nsec.
>> */
>> if (ssl_ctx->crl_last_mtime.tv_sec <= crl_stat.st_mtime)
>> {
>> ssl_ctx->crl_last_mtime.tv_sec = crl_stat.st_mtime;
>> backend_tls_ctx_reload_crl (ssl_ctx, crl_file, crl_file_inline);
>> }
>>
>> I slightly prefer this over adding the extra argument, but can live with
>> your approach just fine too. Pick what you like best :)
>
> I guess the assumption here is that crl_last_mtime.tv_sec is initialized with
> 0.
> This would make your code work. And yeah, I also prefer to avoid the reload
> argument ;)
Correct. This might not be immediately clear in the code, but we have
the good habit of zero-initilializing structs like this :)
> Will make these changes and resend the patch.
Thanks!
-Steffan
------------------------------------------------------------------------------
_______________________________________________
Openvpn-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/openvpn-devel
