Re: New Package: sticky-shell

[Andrew De Angelis <bobodeangelis@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 4445 bytes --]

@Jean Louis, about

> I have tested it, it does not work reliably.
>
Thanks for flagging this. What version of Emacs are you running, and do you
have any special shell configuration?
I built Emacs from master last week, did "emacs -q" and sticky-shell works
as expected.

@Stefan Monnier

> Any chance we could simply add it to `shell.el`?
>

I wouldn't mind adding it to `shell.el`. Let me know what the process would
be in that case.

@Philip Kaludercic

> I assume you are referring to GNU ELPA (the only where contributors have
> to sign the FSF CA?)
>
Yes.

And thanks for your suggestions/comments on the code. I will apply the
changes and push them soon,  including adding the .elpaignore file.

I do have some questions about some of your comments:

+    ;; Perhaps you should pull this into a separate function, as the
+    ;; check appears quite often.  Another idea, as the pattern appease
quite similar in general, you could also
I think part of this comment is missing? But yes I do agree a separate
function is probably a good idea here.

@@ -129,23 +134,25 @@ macro-expands to:
  (upcase
   (funcall sticky-shell-get-prompt))
  \\='face \\='minibuffer-prompt)"
+  ;; The case distinction appears unnecessary (thread-first (foo)) is
+  ;; the same as (foo).
   (if sticky-shell-prompt-modifiers
       `(thread-first
          (funcall sticky-shell-get-prompt)
          ,@sticky-shell-prompt-modifiers)
+    ;; Perhaps it would be better/cleaner if
+    ;; `sticky-shell-prompt-modifiers' were a list of function that
+    ;; all get applied on the result of (funcall
+    ;; sticky-shell-get-prompt) in order?
     (funcall sticky-shell-get-prompt)))
Not sure what you mean by the first comment.
Regarding the second comment: this is essentially what thread-first does,
with the added advantage that `sticky-shell-prompt-modifiers' can consist
not only of functions, but also of forms with multiple arguments, and whose
first argument will be the result of (funcall sticky-shell-get-prompt) at
runtime. For example, if users wanted to `propertize' the header with
particular properties, they wouldn't be able to do it if
`sticky-shell-prompt-modifiers' was simply a list of functions operating on
a single argument, because `propertize' requires multiple arguments. The
solution would have to be to create a new function:
(defun propertize-my-way (string)
         (propertize string 'face 'minibuffer))
And add propertize-my-way to the list of prompt modifiers.
Using thread-first saves users time by allowing them to simply add
'(propertize 'face 'minibuffer) to `sticky-shell-prompt-modifiers'.

On Mon, Dec 12, 2022 at 10:27 PM Jean Louis <bugs@gnu.support> wrote:

> * Andrew De Angelis <bobodeangelis@gmail.com> [2022-12-12 06:24]:
> > This is the git repo: https://github.com/andyjda/sticky-shell, with
> > additional info in the README and in the code's documentation.
>
> Just that the "sticky shell" does not reflect what it does.
>
> I have tested it, it does not work reliably.
>
> - it does not show which command generated error, I have for example:
>
> $ youtube-dl -f18 "https://www.youtube.com/watch?v=PVsnbalE6Ig"
>   C-c C-cTraceback (most recent call last):
>   File
> "/usr/lib/python3.10/site-packages/youtube_dl/extractor/__init__.py", line
> 4, in <module>
>     from .lazy_extractors import *
> ModuleNotFoundError: No module named 'youtube_dl.extractor.lazy_extractors'
> ...with more lines here below...
>
> and that one is not shown.
>
> - for this below it shows "dcon" instead of "dconf":
>
> $ dconf
> error: no command specified
>
> Usage:
>   dconf COMMAND [ARGS...]
>
> Commands:
>   help              Show this information
>   read              Read the value of a key
>   list              List the contents of a dir
>   write             Change the value of a key
>   reset             Reset the value of a key or dir
>   compile           Compile a binary database from keyfiles
>   update            Update the system databases
>   watch             Watch a path for changes
>   dump              Dump an entire subpath to stdout
>   load              Populate a subpath from stdin
>
> Use 'dconf help COMMAND' to get detailed help.
>
>
> - for ls -lR it shows "ls -l"
>
> I have got feeling that it works unreliably.
>
> --
> Jean
>
> Take action in Free Software Foundation campaigns:
> https://www.fsf.org/campaigns
>
> In support of Richard M. Stallman
> https://stallmansupport.org/
>

[-- Attachment #2: Type: text/html, Size: 6487 bytes --]