Re: [evolution-patches] [Fwd: [composer, mailer] #59490 Signature is prefixed with '> " in replies]

[Radek Doulík <rodo novell com>]

On Mon, 2004-07-26 at 15:28 +0800, Not Zed wrote:
>     
>     A couple of minor things.
>     
>     If you're going to compare a size_t against -1 you should use ~0, or alternatively make it a ssize_t as size_t isn't signed.
OK, I used ssize_t

>     And this bit looks abiguous, since arguments are normally pushed onto the stack right to left, and ',' expressions are evaluated right to left.
>     
>     +               e_msg_composer_set_pending_body (composer, em_utils_part_to_html (mime_part, &len), len);
>     
>     e.g. put len (undefined) on the stack, then evaluate something that calculates it.
>     
>     At least thats the way i remember the language working.  But to make it clearly unambiguous i'd suggest separating out the statements, even if it actually works.
Yeah, you are right. For some reason I thought that it will first call all the sub-calls and then push everything on the stack (I think I spent too much time with ppc ;). It indeed mix push to stack with subcalls on i386, at least with -O0. It works differently when optimizations are on. So I made it call em_utils_part_to_html first as you suggested.

>     Otherwise looks good.
OK. I've committed it with the changes to CVS.

Cheers
Radek