[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: master babe6a5e948 1/2: Introduce a new TRAMP method `androidsu'
From: |
Michael Albinus |
Subject: |
Re: master babe6a5e948 1/2: Introduce a new TRAMP method `androidsu' |
Date: |
Wed, 28 Feb 2024 16:38:15 +0100 |
User-agent: |
Gnus/5.13 (Gnus v5.13) |
Po Lu <luangruo@yahoo.com> writes:
Hi,
> Michael Albinus <michael.albinus@gmx.de> writes:
>
>> FWIW, after a very very first look on your code I'm even not convinced
>> we need a new Tramp backend for what you want to achieve.
>
> Perhaps not, but at the very minimum a replacement for:
>
> ;; Android su binaries contact a background service to
> ;; obtain authentication; during this process, input
> ;; received is discarded, so input cannot be
> ;; guaranteed to reach the root shell until its prompt
> ;; is displayed.
> (with-current-buffer (process-buffer p)
> (tramp-wait-for-regexp p tramp-connection-timeout
> "#[[:space:]]*$"))
>
> will be required during connection setup. Thanks.
I will check the majority of the changes over the weekend. I'll be OOO
next two days.
For now, I've reviewed your changes in tramp-adb.el. In general they are
OK, except the check for tramp-androidsu-method in tramp-adb-send-command.
It doesn't make sense (we must special case also tramp-adb-method), so
I've reverted this. In general, even if we're going with your approach,
we must NOT produce spaghetti code. Using tramp-adb objects in
tramp-androidsu might be OK, but then not the other direction.
Best regards, Michael.