This still always decrypts the password, because `ssh-add` does not check
whether the key is already present in the current agent. This should propably
done separately before calling `ssh-add`.
This is to prepare computing the password only when needed, i.e., when the key
is not already included in the running agent. We are not there yet, though.
I am not sure why it hasn't been working until now, and I am not sure why it
indeed does work now, but it does, and that sufficient for me – at least for
now.
The original version of `grep-read-files` includes file names in its default
values, giving an irritating completion candidate list when used with ivy.
Changed this to just let `completing-read` do the completion itself.
Using `insert` directly makes use of character conversion and may scramble the
byte when inserting into the buffer (indeed, `insert` does not seem to insert
the byte, but the characters whose code-points is given in the string; same for
`insert-char`).
When listing files, it's not relevant whether the file is readable or whether
a symbolic link points to a non-existing file. What matters is that the file
itself exists, either as a file or as a symbolic link.
This is relevant when using `git annex find` to list files matching some search
criteria.
This allows easier updates of this list, without having resort to executing the
corresponding code manually. In the future, we could even update that list
automatically by attaching the new function to some of projectile's hooks or
something.
Outlook seems to expect CRLF in S/MIME signed+encrypted mails, so we add those
somewhere in the process of encoding the mail. Furthermore, Outlook is sending
MIME messages with CRLF line endings, and we have to take care of that when
looking for the end headers.
The changes proposed here are preliminary and subject to further testing.
Flycheck used to be activated unconditionally, resulting in annoying warnings
when evaluating Lisp expressions in the minibuffer with `pp-eval-expression`,
and in the scratch buffer.
In dired, offer all marked files, or the currently selected file if there are
none. Otherwise, just use the list of recently opened files as completing.
This is meant as a very simple replacement for dired+'s dired bookmarks.
It's a bit clearer now what the function is doing when ARG is given, I hope. It
also turns out that switching to the current working directory does not make
much sense when we are in the shell buffer, because CWD is then just the, well,
current directory. The original logic used to CWD of the previous buffer (by
closing the shell buffer and immediately reopening it), but that's actually not
what the function is supposed to be doing, is it?
A curcial step in the conversion, namly the computation of the actual time from
the seconds since the epoch, had been conducted with too little precision. Now
the precision is fixed to a high value throughout the whole computation, and the
tests succeed again.