> We tried to make the error values in sockets.c
standard conformant,
> so that's probably the reason they changed from 1.3.2 to 1.4.1. Do
> you think there is an error in the compatiblity or do you just want
> to get the old values back? Yeah you're right. EALREADY is what connect() shall
return when being called again and no connection is established yet. It
just looks awkward to always translate ERR_ISCONN to EALREADY just like
that. Why not rename ERR_ISCONN to ERR_ALREADY and change
the comment while at it. That way, the translation table wouldn't look
suspicious and the usage in api_msg.c would look more correct according
to the standard.
In the final behavior, LwIP is probably closer to
"the standard" now. Still my guys are complaining about their programs
so we might keep it at EISCONN...
> BTW: The "this would be how I'd like it"
line below seems wrong,
> since an lwip error is in the table... Yeah. I copy&pasted it and changed it in the mail
without thinking. Sorry.
> BTW2: EALREADY is defined in arch.h since 2003...
Did you mix up
> standard error defines with lwip error defines? No my IDE just took the errnos out of the index because
I unset LWIP_PROVIDE_ERRNO,
so it didn't show up in my search. My mistake. Sorry again.