Dumbest. Problem. EVER.
Sep. 26th, 2008 12:55 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
theweaselkingCVS server set up. User connects, remotely, via SSH to the server to do servery-things and also via TortoiseCVS to make CVS checkouts.
User's account worked in the past, with the same software, same PC, same network, etc.
User's settings in Tortoise are correct.
User's account works for me - same software, etc.
User can connect to the server via PuTTY, no problem, using *exactly* the same username, password, and port that Tortoise uses.
However, user's password does not work when his computer's copy of Tortoise submits it. He has reinstalled the program, still has the same problem.
The only weird thing: When you get a password error from Tortoise, it doesn't say that user@servername's password is wrong, is says that @servername's password is wrong - no "user". Despite the fact that Tortoise's settings show that it is submitting username and password (same as my PC), and the CVSROOT showing :ssh:user@servername, and the *password prompt* knowing the right username, the error message shows that no username is being submitted.
The source of this problem, which is the DUMBEST THING EVER?
Tortoise uses PLink to make a connection.
Plink takes two parameters - username and host.
"host" can be either a hostname, *or the name of a connection entry in PuTTY*.
This user, when connecting through PuTTY, had set the name of his saved connection settings to be the same as the name of the server.
Rename that *label* inside *a different program* so that now the *label* says "Companyname server" rather than "server.domain.tld", and suddenly Tortoise's personal, private, hidden-and-included copy of PLink uses the correct parameters it's been passed instead of just kind of guessing.
HATE ALL PROGRAMMERS.
User's account worked in the past, with the same software, same PC, same network, etc.
User's settings in Tortoise are correct.
User's account works for me - same software, etc.
User can connect to the server via PuTTY, no problem, using *exactly* the same username, password, and port that Tortoise uses.
However, user's password does not work when his computer's copy of Tortoise submits it. He has reinstalled the program, still has the same problem.
The only weird thing: When you get a password error from Tortoise, it doesn't say that user@servername's password is wrong, is says that @servername's password is wrong - no "user". Despite the fact that Tortoise's settings show that it is submitting username and password (same as my PC), and the CVSROOT showing :ssh:user@servername, and the *password prompt* knowing the right username, the error message shows that no username is being submitted.
The source of this problem, which is the DUMBEST THING EVER?
Tortoise uses PLink to make a connection.
Plink takes two parameters - username and host.
"host" can be either a hostname, *or the name of a connection entry in PuTTY*.
This user, when connecting through PuTTY, had set the name of his saved connection settings to be the same as the name of the server.
Rename that *label* inside *a different program* so that now the *label* says "Companyname server" rather than "server.domain.tld", and suddenly Tortoise's personal, private, hidden-and-included copy of PLink uses the correct parameters it's been passed instead of just kind of guessing.
HATE ALL PROGRAMMERS.
(no subject)
Date: 2008-09-26 05:13 pm (UTC)Sorry, already followed this directive long before you issued it.
Diagnostic skills
Date: 2008-09-26 05:49 pm (UTC)But they wouldn't hate Tortoise's developer; they might even blame themselves. So this is what your diagnostic skills really get you: the ability to assign blame correctly to others. Given that that, why hate all programmers? Hate the ones who did this, instead. You've earned it (and so have they).
By the way, I still think you should replace Tortoise. Anyone who did this to you is doing something else to you *right now*. You just don't know it yet.
Re: Diagnostic skills
Date: 2008-09-26 06:00 pm (UTC)And CVS clients other than Tortoise?
REALLY suck.
anyone without a sense of the absurd would never even entertain the notion that changing that might fix the problem.
I got to it by going step by step.
Password fails from Tortoise -> Password works from Putty -> same password, port, and server, so something's wrong with Tortoise -> Tortoise settings match those of a known-good working installation perfectly -> reinstall of Tortoise fixes nothing -> Wait, looks like Tortoise is sending the wrong *username*, not just bad password -> Manually run Tortoise's SSH connector, from the command line, in verbose mode, get same password error for same reason -> check the usage listing, note that line about how "host" can be either a hostname or a PuTTY configuration -> change the name of the PuTTY configuration so it's not the same as the hostname -> problem solved.
This just takes a willingness to step through the problem, narrowing down where the problem had to be each time.
By the way, I still think you should replace Tortoise. Anyone who did this to you is doing something else to you *right now*. You just don't know it yet.
Best of a bad lot, and it's been working perfectly for this bunch for *years* until this problem. All other CVS clients have caused more problems, sooner.
Re: Diagnostic skills
Date: 2008-09-26 06:16 pm (UTC)"Best of a bad lot"; okay, I buy that other pieces of software might be doing *more* to you.
Re: Diagnostic skills
Date: 2008-09-26 06:26 pm (UTC)And yes, I was almost at the point of switching to WinCVS or something similar - I'd confirmed that SSH wasn't the problem, after all, and WinCVS would either work (and confirm that Tortoise was truly gefucked) or fail (and confirm that there's something wrong with *cvs* from that machine, not just Tortoise.)
My client would not have been happy, though. He *likes* Tortoise, just like I do, which is why I spent the time to look through it.
Re: Diagnostic skills
Date: 2008-09-26 06:27 pm (UTC)Re: Diagnostic skills
Date: 2008-09-26 06:50 pm (UTC)Might?
The only explanation for the state of software today is that programmers are either malicious or incompetent, but John's not kidding when he states that all software sucks. It's only a matter of degree.