Michael Albinus <[email protected]> writes:

> Eshel Yaron <[email protected]> writes:
>
>> Hi Michael,
>
> Hi Eshel,
>
>> Thanks!  It makes sense and I think we could make it work.
>
> I will check where we need changes. And documentation. I don't know
> whether it will be ready for Tramp 2.8.1.3, which is already tested and
> scheduled for release end of this month. Let's see.

Great, and no pressure for the upcoming release.

>> My only concern is this: IIUC, tramp-connection-properties is meant for
>> users to override tramp-methods parameters that were defined by the
>> method author, but here we'd be (ab)using tramp-connection-properties
>> as part of the method definition.
>
> That's not an abuse. Overwriting tramp-login-program is a common (and
> documented, I believe) use case. Another use case is to overwrite
> tramp-connection-timeout for sudo-like methods. And in general it can be
> used to tweak tramp-login-args. People do so.

The people who do so are users, right?  
My point was that here we'd be relying on tramp-connection-properties as
part of the method definition, not as something that users configure.
Namely, when users will run kubed-tramp-enable to enable the kubedv2
method, it'll modify tramp-connection-properties at the same time as it
adds the method to tramp-methods.  (Note that this means that the
tramp-login-program in tramp-methods is never actually used.)
Are there any packages/methods that modify tramp-connection-properties?
I searched a bit and couldn't find any.

> [...]
>
>> (The mismatch is also apparent in the
>> fact that we use a regexp that only looks at the method name.)  In that
>> sense, ISTM that a similar generalization of tramp-methods would be a
>> little nicer.
>
> But this *is* a generalization of tramp-methods, isn't it?

What I meant was allowing tramp-login-program in tramp-methods to be a
function, instead of doing that in tramp-connection-properties.  
That way we wouldn't have to touch tramp-connection-properties in the
method definition, just tramp-methods.

>> But again, in practice, I think it could get the job done.
>> WDYT?
>
> You might check in parallel how it works in your case. Please ping me
> when there is a problem.

Will do!


Thanks,

Eshel

Reply via email to