TCPIP buffer problem?

[Uwe Zessin]

It's been some time I played with a C compiler, but there are functions like:

http://www.opengroup.org/onlinepubs/009695399/functions/strerror.html
http://www.opengroup.org/onlinepubs/009695399/functions/perror.html

[Willem Grooters]

(this tends to get off-topic....)

Uwe,

Thanks for these links, but getting the message string isn't the real point. I would like to be able to do something like:

if lib$machc_cond (errval, TCPIP_WOULDBLOCK)

that requires errval to contain a full, VMS-like returnstatus (32 bit, format: Facility/Number/Severity.
Number could well be 35 - the value of errno, being EWOULDBLOCK.

Of course I could do

if (errno == EWOULDBLOCK)

but I'm programming on VMS, not Unix!

(Like a lot of things in CRTL, the behaviour could be controlled with some DECC$-logical)

OpenGroup = Unix. They should be looking more widely!

Willem

[Uwe Zessin]

I doubt something like the attached make you more happy. It is just numbers and the errno value does not have a severity anyway, so you can even save a function call by doing a direct compare against EWOULDBLOCK.

[Willem Grooters]

Uwe,

Suppose the program used SYS$QIO in stead of plain Unix-like socket access. I doubt it would return '35' in IOSB for cases like this. No doubt there is a plain VMS alternative.
OTOH - since socket programming IS unix-like (VMS _would_ use SYS$QIO instead) there is something to say to do a direct compare, the Unix way. Why not either one OR the other, the same way CTRL is taking care...

For this thread, it's not the point. The problem IS solved another way, and any error IS an error. send should succeed.

Willem