Re: [PATCH] Re: [fetchmail-devel] Re: 6.2.6 fails where 6.2.5 was OK

[David G. <da...@dg...>]

Matthias Andree wrote:

> You've got the patch backwards. Use "diff -ur OLDER NEWER".

sorry.

>Even reading it in the right direction, I don't see what this does - and
>how it could possibly fix your problem. use_ssl defaults to false,
>
use_ssl *should* default to false.

I didn't follow the logic through - it was an eyeballing of the use of 
an _ssl variable outside an SSL_ENABLE block.
It does actually fix the problem for me.

Once an 'obvious' bug fixed it for me I passed it on to you...
Anyway, I'm glad it's helped find a regression.

This code snippet was my debug test... I didn't know the insides of 
fetchmail well enough to assume anything else :)


    if (ctl->use_ssl == FLAG_FALSE) fprintf(stderr, "is FLAG_false\n");
    else fprintf(stderr, "use_ssl isn't FLAG_false\n");
   
    if (
#ifdef SSL_ENABLE
        (ctl->use_ssl != FLAG_FALSE) ||
#endif /* SSL_ENABLE (*/


fetchmail: 6.2.5.991 querying pop.freeserve.net (protocol POP3) at Sat 
19 Mar 2005 01:07:16 PMGMT: poll started
fetchmail: POP3< +OK connected to pop3 on 3002
use_ssl isn't FLAG_false
fetchmail: POP3> CAPA
fetchmail: POP3< -ERR unknown command
fetchmail: unknown command
fetchmail: Repoll immediately on 
da...@in...@pop.freeserve.com
fetchmail: POP3< +OK connected to pop3 on 3212
use_ssl isn't FLAG_false
fetchmail: POP3> USER da...@in...
fetchmail: POP3< +OK name is a valid mailbox
fetchmail: POP3> PASS *
fetchmail: POP3< +OK user exist with that password

################################################
anyway.... I couldn't leave it at that :)


a context grep of use_ssl gives an interesting result in fetchmail.c:
fetchmail.c-#ifdef      SSL_ENABLE
fetchmail.c:        DEFAULT(ctl->use_ssl, FALSE);
fetchmail.c-        DEFAULT(ctl->sslcertck, FALSE);
fetchmail.c-#endif

so if SSL_ENABLE is not defined it's undefined.

So, I'm thinking my patch was right after all ;)

:-P  nyah!

David