I apologize if I use the incorrect terminology here, I've only been using emacs for a few months.
I just reinstalled Aquamacs on a macbook that I reformatted and am having the oddest problem.
I open a .py file, and use C-c ! to open a python shell. So I have (as expected), the .py file in the top window and the python shell in the bottom window.
If I then run C-c C-c (py-execute-buffer) in the .py file, the two windows swap positions. I mean, the .py file buffer opens in the bottom window in a new buffer, while the python shell opens in the top window in a new buffer. So basically, they swap positions. Repeatedly using C-c C-c swaps the windows back again... so they're shuffling positions. Also, both windows (top and bottom) have both buffers (.py file and python shell) in tabs.
I've not made any modifications to the default settings yet, and I've gotten the problem with both 2.3a and 2.3 (2.3 was on the machine previously and didn't have this problem, so I tried rolling back... to no avail).
Does anyone know how to stop this behavior? Thanks in advance!
Add the following to your Emacs init file in Aquamacs to prevent it from swapping the buffers around:
(defadvice py-execute-buffer
(around keep-buffers-same activate)
"Don't swap buffers in Aquamacs."
(save-window-excursion
ad-do-it))
You can also try adding the following to your emacs init file:
(setq py-split-windows-on-execute-p nil)
This will prevent the current window from splitting after you run any py-execute-*. (This also means that the python shell won't show up if it isn't already in one of your windows.)
i don't use Aquamacs and couldn't reproduce your described behaviour, however, try this code to toggle either of the windows as 'dedicated'. locking windows to buffers was the first thing i wanted to do when getting up and running with emacs. maybe this will help you here.
add the code to your '.emacs', then either 'mark' (select) the region 'S-< key-down >' and then 'M-x eval-region' to evaluate it ..or save and restart emacs.
(global-set-key [pause] 'window-dedication-toggle)
(defun window-dedication-toggle (&optional window force quiet)
"toggle or ensure the 'dedicated' state for a window"
(interactive)
(let* ((toggle t) (window (if window window (selected-window))) (dedicated (window-dedicated-p window)))
(cond ((equal force "on") (setq toggle (eq dedicated nil)))
((equal force "off") (setq toggle (eq dedicated t))))
(if toggle (progn
(setq dedicated (not dedicated))
(set-window-dedicated-p window dedicated)
(if (not quiet)
(message "window %sdedicated to %s" (if (not dedicated) "no longer " "") (buffer-name)))))))
Related
I'm fairly new to Emacs and finally figured out how to set up python environment. I'm using elpy and iPython as python shell interpreter. My question is as follows:
After I press C-c C-c, the buffer on the right shows the executed result. The red box is the code being executed and the blue box is the executed result returned from python. My questions are:
Is there a way only showing the result?
And every time after the script being executed, how to set the buffer on the right only showing the current result, ie, clear previous executed results? Thanks.
Is there a way only showing the result?
elpy already has a custom variable for this:
(setq elpy-shell-echo-input nil)
And every time after the script being executed, how to set the buffer
on the right only showing the current result, ie, clear previous
executed results? Thanks.
elpy doesn't have this functionality. But you can create a wrapper function and bind it to the same keys:
(define-key elpy-mode-map (kbd "C-c C-c")
(lambda (p)
(interactive "P")
(ignore-errors
(with-current-buffer (process-buffer (python-shell-get-process-or-error))
(let ((comint-buffer-maximum-size 0))
(comint-clear-buffer))))
(elpy-shell-send-region-or-buffer p)))
My .emacs file (emacs 23.4.1) contains python and latex related code. For both there is eval-after-load (code that I want to be executed just once when emacs initiates) and hooks. A relevant part of it is:
(setq py-install-directory "~/.emacs.d/python-mode.el-6.1.3")
(add-to-list 'load-path py-install-directory)
(require 'python-mode)
(defun my-eval-after-load-python()
(setq initial-frame-alist '((top . 48) (left . 45) (width . 142) (height . 57)))
(split-window-horizontally (floor (* 0.49 (window-width)))))
(eval-after-load "python-mode" '(my-eval-after-load-python))
All hooks work fine, but my-eval-after-load-python doesn't, which causes the frame to be split into two windows everytime emacs initiates for every extension (for example: emacs file.py, emacs file.tex, emacs file). I tried to change it to:
(eval-after-load "python-mode"
'(progn
(setq initial-frame-alist '((top . 48) (left . 45) (width . 142) (height . 57)))
(split-window-horizontally (floor (* 0.49 (window-width))))
, but it still doesn't work. There's probably a beginner mistake going on here, but I'm unable to find it. How would I split the window just the first time a python script is opened (emacs file.py) and not every time I open a new buffer file2.py?
It sounds like something is causing (load "python-mode") to happen "everytime emacs initiates for every extension" (I'm not sure what you actually mean by that).
Your code is also strange in that you are forcibly loading python-mode with require, and then subsequently evaluating eval-after-load for that same library, even though you know that it's definitely already loaded. This will still work, but it's odd. One tends to use eval-after-load to avoid loading something up front (letting autoloading deal with it on demand, but still having the custom code running at that time).
Edit: Oh, do you just mean that when you start Emacs it evaluates your eval-after-load code? That's because you've told it to -- you loaded python mode, and then told Emacs that if/when python mode is loaded, split the screen in two.
Maybe introduce some boolean
(defvar my-action-done nil)
Put them non-nil following action and
(unless my-action-done
do ...
You do (require 'python-mode) right at the beginning, so python-mode is always loaded even before you get to the (eval-after-load "python-mode" ...) part. I thinkg your my-eval-after-load-python is not meant to be loaded when python-mode is loaded but when python-mode is entered. So you want to use
(add-hook 'python-mode-hook #'my-eval-after-load-python)
When I am debugging code in Emacs, I like to have two buffers open, the first one with the source code I am debugging, and the second one with the debugger (pdb for Python).
I have the following keyboard shortcuts defined on my .emacs file:
(require 'gud)
(define-key gud-mode-map '[C-f10] 'gud-next)
(define-key gud-mode-map '[C-f11] 'gud-step)
(define-key gud-mode-map '[C-f5] 'gud-cont)
(define-key gud-mode-map '[C-f12] 'gud-break)
With the above, I can trigger the GUD shortcuts for gud-next, gud-step, etc. from the buffer where pdb is running, but I can't trigger them from the buffer that has the python code.
I would like to use keyboard shortcuts on the buffer with the source code to trigger GUD commands for the debugger. Is there any way to do this?
I am using the most recent version of python-mode (6.0.4) and Emacs 23.3.1.
Try using global-set-key instead:
(global-set-key [C-f10] 'gud-next)
(global-set-key [C-f11] 'gud-step)
(global-set-key [C-f5] 'gud-cont)
(global-set-key [C-f12] 'gud-break)
IIRC, this worked for me.
I am a fairly proficient vim user, but friends of mine told me so much good stuff about emacs that I decided to give it a try -- especially after finding about the aptly-named evil mode...
Anyways, I am currently working on a python script that requires user input (a subclass of cmd.Cmd). In vim, if I wanted to try it, I could simply do :!python % and then could interact with my script, until it quits. In emacs, I tried M-! python script.py, which would indeed run the script in a separate buffer, but then RETURNs seems not to be sent back to the script, but are caught by the emacs buffer instead. I also tried to have a look at python-mode's C-c C-c, but this runs the script in some temporary directory, whereas I just want to run it in (pwd).
So, is there any canonical way of doing that?
I don't know about canonical, but if I needed to interact with a script I'd do M-xshellRET and run the script from there.
There's also M-xterminal-emulator for more serious terminal emulation, not just shell stuff.
I like to use the Emacs "compile" command to test/run my python scripts. M-XcompileRET will pull up the default "make -k" but if you delete that and put in the command line for your script (including options), subsequent "compiles" will provide the new "compile" command automatically. All the output from your script will appear in the compile buffer. (As opposed to the shell, this provides a nice clean buffer each time it is invoked. Good for searching and such. If you forget to save your script before your run, compile will ask you if you would like to save the file.)
You will lose your the command line when you restart Emacs. But you can get Emacs to set the compile-command for the buffer holding your script by putting at the bottom of the python script this sort of code (actually a python comment):
# Trigger emacs to run this script using the "compile" command
# ;;; Local Variables: ***
# ;;; compile-command: "my_cool_script.py --complicated_option some_filename.txt" ***
# ;;; end: ***
This is handy for scripts with complicated invocations.
Note: The python comment character '#' protects this from the python interpreter while Emacs knows to set these variables because it looks at the bottom of every file when it opens them.
I'd love to be able to jump to 'compile errors' in my python script the way the compile command does when you use it for compiling C code but I'm too lazy to create the Emacs regular expression to make this work. Perhaps that would make another great question for stack overflow!
I currently use these hook to define my compilation commands:
(defun convert-filename-to-executable (file)
(if (eq system-type 'windows-nt)
(concat (file-name-sans-extension file) ".exe")
;; linux
(concat "./" (file-name-sans-extension file))))
(add-hook 'c++-mode-hook
(lambda ()
(unless (file-exists-p "Makefile")
(set (make-local-variable 'compile-command)
(let* ((file (file-name-nondirectory buffer- file-name))
(executable (convert-filename-to-executable file)))
(concat "g++ -g -Wall -o "
(file-name-sans-extension file)
" "
file
" && "
executable))))))
(add-hook 'c-mode-hook
(lambda ()
(unless (file-exists-p "Makefile")
(set (make-local-variable 'compile-command)
(let* ((file (file-name-nondirectory buffer-file-name))
(executable (convert-filename-to-executable file)))
(concat "gcc -g -ansi -Wall -Wpedantic -Wextra -Wc++-compat -Wconversion -o "
(file-name-sans-extension file)
" "
file
" && "
executable))))))
(add-hook 'python-mode-hook
(lambda ()
(set (make-local-variable 'compile-command)
(concat "python " buffer-file-name))))
(add-hook 'perl-mode-hook
(lambda ()
(set (make-local-variable 'compile-command)
(concat "python " buffer-file-name))))
Together with this lambda set to call the compile function interactively:
(global-set-key (kbd "<f4>") (lambda () (interactive) (setq current-prefix-arg '(4)) (call-interactively 'compile)))
One button to rule them all!
If you press F4 (in my case, you can set the key yourself in the lambda for global-set-key), then a file opened in C++ or C mode will be compiled and a file in python or perl mode will be run (interactively)
The I think ansi-term has the most faithful emulation of a terminal. But I don't see a way to pass arguments to the process. You can of course just launch it from a shell inside the ansi-term buffer.
But I think the best thing to do is to not use python-send-buffer, but instead to use a new function which does it "right", that is by sending the path to the current file instead of making a temp file. There are some slight differences of course in that you have to save the current file first, but the following should at least get you on the right track.
(defun python-send-file ()
(interactive)
(save-buffer)
(python-send-string (concat "execfile('" (buffer-file-name) "')")))
;; This overwrites the `python-send-buffer' binding so you may want to pick another key
(eval-after-load "python"
(define-key python-mode-map "\C-c\C-c" 'python-send-file))
I checked and this allows you to interact. To get tabs you have a few options.
C-qTAB will always give you a literal tab
You can rebind tab to be a literal tab in inferior-python-mode-map:
(define-key inferior-python-mode-map "\C-i" 'self-insert-command)
I'm sure there are others that I can't think of
If you use C-c C-c a buffer is created (look for inferior-python). Try changing to that buffer*, every time you hit C-c C-c the result is shown there, you need to see that buffer to get the results. Use C-x 2 so you can see both buffers at the same time.
Also try C-c C-z (switch to shell).
*I use Ibuffer to manage buffers, is very good.
(btw, this http://tuhdo.github.io/index.html is an excelent place to learn some emacs)
EDIT: Have you tried C-c C-p ?
Yet another option:
Using C-c C-c works nicely with fgallina's python.el -- pwd will be the location of the buffer's file.
I'm using emacs23 with tramp to modify python scripts on a remote host.
I found that when I start the python shell within emacs it starts up
python on the remote host.
My problem is that when I then try to call python-send-buffer via C-c C-c it comes up with the error
Traceback (most recent call last):
File "", line 1, in ?
ImportError: No module named emacs
Traceback (most recent call last):
File "", line 1, in ?
NameError: name 'emacs' is not defined
Now, I must admit that I don't really know what's going on here. Is there a way for me to configure emacs so that I can evaluate the buffer on the remote host?
Many thanks.
Edit: I've followed eichin's advice and re-implemented python-send-region. See my answer below.
I'm currently trying to to get my unregistered question merged with this account, after which I'll be able to accept eichin's answer and edit my post to include my solution.
I followed eichin's suggestion and copied the emacs2.py emacs3.py and emacs.py files to the remote host and added their directory to PYTHONPATH in the tramp-remote-process-environment variable.
I then reimplemented the python-send-buffer function in my .emacs
(require 'python)
(defun python-send-region (start end)
"Send the region to the inferior Python process."
(interactive "r")
(let* ((loc_name)
(f (if (file-remote-p default-directory)
(let* ((con (tramp-dissect-file-name default-directory)))
(setq loc_name (tramp-make-tramp-temp-file con))
(concat "/"
(tramp-file-name-method con) ":"
(tramp-file-name-user con) "#"
(tramp-file-name-host con) ":"
loc_name
))
(setq loc_name (make-temp-file "py"))))
(command (format "emacs.eexecfile(%S)" loc_name))
(orig-start (copy-marker start)))
(save-excursion
(let ((curbuf (current-buffer))
(tempbuf (get-buffer-create "*python_temp*")))
(set-buffer tempbuf)
(delete-region (point-min) (point-max))
(insert-buffer-substring curbuf start end)
(python-mode)
(when (save-excursion
(goto-char (point-min))
(/= 0 (current-indentation)))
(python-shift-left (point-min) (point-max)))
(write-region nil nil f nil 'nomsg))
(python-send-command command)
(with-current-buffer (process-buffer (python-proc))
;; Tell compile.el to redirect error locations in file `f' to
;; positions past marker `orig-start'. It has to be done *after*
;; `python-send-command''s call to `compilation-forget-errors'.
(compilation-fake-loc orig-start f)))
))
I essentially copy the region into a new buffer, adjust the indentation and then write it into a temporary file, created with tramp-make-tramp-temp-file or make-temp-file, depending on whether the visited file is remote or local.
I had some problems with tramp-handle-write-region, which didn't seem to accept a string as a first argument, which is why I did all the formatting in a separate buffer first.
Let me know if there are still any problems with the code, but this is my first attempt at elisp coding, so please be gentle.
Short answer: not without writing some missing elisp code.
Long version: In python.el, run-python adds data-directory (which on my Ubuntu 10.10 box is /usr/share/emacs/23.1/etc/ ) to $PYTHONPATH, specifically so that it can find emacs.py (as supplied by the local emacs distribution.) Then it does a (python-send-string "import emacs") and expects it to work...
It looks like the defadvice wrappers that tramp uses don't actually pass PYTHONPATH, so this doesn't work even if you have the matching emacs version on the remote system.
If you M-x customize-variable RET tramp-remote-process-environment RET
then hit one of the INS buttons and add PYTHONPATH=/usr/share/emacs/23.1/etc then hit STATE and set it to "current session" (just to test it, or "save for future sessions" if it works for you) it almost works - the complaint goes away, in any case, because the remote python can now find the remote emacs.py. If you now go back to the original question, doing python-send-buffer, you just run into a different error: No such file or directory: '/tmp/py24574XdA' because python-mode just stuffs the content into a temporary file and tells the python subprocess to load that.
You'd have to change python-send-region (the other functions call it) and particularly the way it uses make-temp-file to be tramp-aware - there's even a tramp-make-tramp-temp-file you could probably build upon. (Be sure to post it if you do...)