On Sun, Nov 04, 2012 at 10:28:41PM +0100, Marcus Hüwe wrote:
Hi Michal,
On 2012-10-31 11:21:41 +0100, Michal Vyskocil wrote:
The biggest problem during usage and testing of a remote build feature is the fact osc asks user several times for the input. In this case, remote process simply hangs and waits for the response, which is not even displayed, because of stdout buffering.
In this case ssh should be invoked with the "-t" parameter.
Corret, I'll update my --host patch instead
The samy apply for non-interactive usage - if osc is called from a script, then it hangs waiting on something on stdin, which will never come.
Therefor I've added a simple check os.isatty(sys.stdin.fileno()) to the osc.core.raw_input() - in case it's not terminal, the new RawInputError is raised. It will show-up the question on stderr and program will be terminated.
We could add something like a "--script-mode" (config) option. If "osc --script-mode <cmd>" is invoked and <cmd> requires a user input an exception is raised.
What do you think?
To be honest, I'd not invest a lot of effort in osc (1.x). Having a reliable scripting mode (IOW not wait on stdin, provide reasonalbe return codes, ...) can be a nice enhancenment of osc2, can't it? Regards Michal Vyskocil