Hi, * Kyle Wheeler wrote: > What I do is simply allow interrupts during socket reads. It's > clunky, because it aborts the connection, but it gets the job done > (see attachment). Well, this doesn't seem too unreasonable. In case errno=EINTR, why not make the socket functions return -1 in general failure and -2 in case of interrupt? That way you can signal interruption as soft error up the hierarchy and do sane things in the downloading code. The same would be useful for aborting SMTP sending, too. Of course, the real fix would be to have a key to background the current network operation and continue working. But that would mean quite a lot of refactoring so we might just want to allow interrupting socket read/write, communicate the interrupt and handle it. Rocco
Attachment:
pgpLs_1WkmHD5.pgp
Description: PGP signature