Giuseppe Scrivano gnu.org> writes:
>
> Indeed. Thanks for your contribution! I have done a trivial change and
> pushed it.
Welcome!
> If you are going to contibute more code to wget, we will need copyright
> assignment to the FSF. If this is the case, please contact me privately
> and I will
Tim Rühsen writes:
> Ok, thanks.
>
> Your patch should go into git.
>
> Looks like, handling other non-fatal events needs some deeper knowledge
> (except GNUTLS_E_INTERRUPTED, which should not occur).
> Whenever the need arises...
Indeed. Thanks for your contribution! I have done a trivial ch
Ok, thanks.
Your patch should go into git.
Looks like, handling other non-fatal events needs some deeper knowledge
(except GNUTLS_E_INTERRUPTED, which should not occur).
Whenever the need arises...
Regards, Tim
Am Montag, 6. Mai 2013 schrieb mancha:
> Hi.
>
> You are right that GNUTLS_E_WARNI
Hi.
You are right that GNUTLS_E_WARNING_ALERT_RECEIVED is not the only
non-fatal return value. In GnuTLS 2.12.x there's
GNUTLS_E_INTERRUPTED,
GNUTLS_E_REHANDSHAKE, GNUTLS_E_WARNING_IA_IPHF_RECEIVED, and
GNUTLS_E_WARNING_IA_FPHF_RECEIVED.
My patch only addresses non-fatal *alerts* (a subset of no
Hi,
thanks for your work to improve wget !
Are you shure, there are no other non-fatal return values ?
e.g. GNUTLS_E_REHANDSHAKE
AFAIK, a GnuTLS example that also uses a handshake loop, but relies completely
on gnutls_error_is_fatal():
// simplified version without timeout handling
do {
Hello.
wget, built against GnuTLS, terminates SSL/TLS handshakes
upon receiving any error alert (including non-fatal ones).
This creates a problem when connecting to servers that support
TLS-SNI and reply with a warning-level unrecognized name alert
(eg. due to misconfiguration).
My patch change