Hi, On Thu, Mar 02, 2017 at 01:35:32PM +0100, Christophe Fergeau wrote: > The code is preexisting Hm, so I should avoid reworking code that does not fit in newer style or standards? > it's arguably error handling Agreed > and the while (TRUE) is equally odd if you ask me (mostly because you > said it would loop at most once anyway). The iteration should happen only once because g_coroutine_socket_wait() does wait G_IO_IN/G_IO_OUT. Its very unlikely that after g_coroutine_socket_wait() returns, we can't read or write. Still, that's not odd. Not odd at all. Unless `goto reread` give us some performance gain here, I can't see goto label being preferred to while(1). In college I would lose valuable points choosing one over the other. > I guess I could live with it with a slightly better changelog, and > maybe a comment indicating this is not really a loop. I can't see a reason to improve the commit log if you don't agree with the changes and I don't really intend to keep discussing over this that seems more a personal preference thing. I might be completely wrong in the end :) Still, IMO I see an improvement while interacting with this code later on. Maybe I should not have sent this to soon but I'm trying to avoid huge patch series whenever I can. See this 'wip-qos' patch in current master [1] and with this series [2]. [1] https://gitlab.com/victortoso/spice-gtk/commit/9c68d7875beafbc49df4dc1b8c58a490fbc355b8?view=parallel [2] https://gitlab.com/victortoso/spice-gtk/commit/326528c9674dbeb765bb50f6d0b729f8a2b45c57?view=parallel Cheers, toso
Attachment:
signature.asc
Description: PGP signature
_______________________________________________ Spice-devel mailing list Spice-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/spice-devel