Stefan Monnier <monnier@iro.umontreal.ca> writes:

Hi,

>> I'm sad to report that blocking timers in TRAMP is not sufficient to avoid
>> it breaking other operations: process filters can also run while TRAMP is
>> sending, and waiting for the result of, a command.
>
> I'm not sure what is the problem with timers and/or process filters
> (e.g. I don't see any process filters in your backtrace), so maybe I'm
> misunderstanding the problem, but FWIW, here's what I noticed:
>
>>   tramp-user-error(nil "Not a Tramp file name: \"%s\"" "/Users/slippycheeze/.emacs.d/network-security.data")
>>   tramp-dissect-file-name("/Users/slippycheeze/.emacs.d/network-security.data")
>>   tramp-vc-file-name-handler(expand-file-name "/Users/slippycheeze/.emacs.d/network-security.data" "/gssh:slippycheeze.X.XXXXXXXX.XXX:")
>
> This indicates that Tramp gets confused and thinks that the
> tramp-vc-file-name-handler was called in response to expand-file-name on
> the file "/Users/slippycheeze/.emacs.d/network-security.data" whereas
> it's called because of the directory argument.
>
> So I believe the (untested) patch below should fix the immediate problem.
> Can you confirm?
>
> Michael?

Yes, the analysis is correct. However, we don't need to check this for
every Tramp invocation. Daniel's reports have been covered always
vc-registered. And indeed, the related `tramp-vc-file-name-handler'
seems to expect always a Tramp file name, which is not true for filters
and sentinels which run for non-Tramp processes.

Daniel, does the appended patch fixes the issue for you?

>         Stefan

Best regards, Michael.