Hi.

I report some concerns about channel_open().

1) https://github.com/vim/vim/blob/0f526f5/src/channel.c#L684 (On OS X)

When "FD_ISSET(sd, &rfds) && FD_ISSET(sd, &wfds)" is satisfied, "ret" is 2, 
thus this if-block is never used.
But removing "ret == 0" (or replacing by "ret == 2"), when we try to connect to 
local port which is not listening, we go into if-block and it returns NULL.
As a result, "waittime" is ignored (unlike other *nix environment), so 
Test_open_delay fails.

2) https://github.com/vim/vim/blob/0f526f5/src/channel.c#L686

Originally, according to the following, this behavior "encounter of errors 
cause a socket to be both readable and writeable" is derived from 4.4BSD 
implementations.

Refs: Unix Networking Programming
https://books.google.co.jp/books?id=ptSC4LpwGA0C&pg=PA449&lpg=PA449&dq=tcp+socket+writable+on+error&source=bl&ots=Kr5DRiimUu&sig=LPaFAJzjQUN4Cmhp-VCLsvVmApM&hl=en&sa=X&ei=EYLcUM3wO9HW0gGb44GABw&redir_esc=y#v=onepage&q=tcp%20socket%20writable%20on%20error&f=false

(and this book mentions that TCP/IP illustrated vol.2 describes about such 
behavior; but I forgot where it is, sorry..)

It appears that other *nix environments behave in the same way.

But not vice versa, so we need check socket error by getsockopt() when socket 
is both readable and writeable after connect().


patch:

https://gist.github.com/ichizok/e68a5b850df8362a0dbb


Thank you.
- Ozaki Kiichi

-- 
-- 
You received this message from the "vim_dev" maillist.
Do not top-post! Type your reply below the text you are replying to.
For more information, visit http://www.vim.org/maillist.php

--- 
You received this message because you are subscribed to the Google Groups 
"vim_dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Raspunde prin e-mail lui