# -*- mode: org -*-
#+TITLE:     TODO
#+SEQ_TODO:  TODO WAITING | DONE CANCELLED MAYBE PARTIAL
#+DRAWERS:   SNIP
#+CATEGORY:  rinari

List of tasks for rinari

for documentation see [[http://rinari.rubyforge.org/]]
for background see http://groups.google.com/group/emacs-on-rails

* TODO Document changing jump keybindings
* TODO Document Xemacs support
* TODO full test coverage (ERT) [0/2]
- [ ] movement tests
  - [ ] controller
  - [ ] view
  - [ ] model
  - [ ] test
  - [ ] rspec
- [ ] compilation tests
  - [ ] console
  - [ ] web-server
  - [ ] test
  - [ ] sql
  - [ ] script

switch from using elunit.el to ert.el for testing

* TODO Merb support [2/4]
  :PROPERTIES:
  :suggested-by: Jorge Calás Lozano
  :END:

** DONE new Merb Root

I think it's better to rely on config/init.rb that file should be on
every merb project, merb.thor doesn't, in fact my "library" project
doesn't have it.

** DONE pull into rinari-merb
rinari-merb (minami? why not just rinari-merb? what if someone create a
rinari version for sinatra? sinasi?)

** TODO load automatically
it doesn't load automatically when you visit a project file or the
project directory (with dired). So I should call minami-minor-mode

** TODO jump support
jump: it doesn't really jump, but seems to look for the correct paths,
at least that is what *Messages* buffer says:

("app/controllers/\\1.rb" "asset")
Searching /home/jorge/dev/jorgecalas/library/app/models/app/controllers/ ...
("app/controllers/\\1.rb" "assets")
("app/controllers/\\1.rb" "asset")
Searching /home/jorge/dev/jorgecalas/library/app/models/app/controllers/
... [2 times]

When I'm on the project directory (dired) it finds files correctly,
right now I tried minami-find-model, minami-find-controller.

minami-find-rspec:

In the current merb you don't don't have spec/controllers or spec/views
(although you can, of course) but spec/requests is encourage. So
app/controllers/assets.rb should map spec/requests/assets_spec.rb

* DONE submit to ELPA
get rinari (and all of it's dependencies) into [[http://tromey.com/elpa/][ELPA]] the Emacs Lisp
Package Archive

* DONE create a screencast
promotion!!

now up with the online documentation [[http://rinari.rubyforge.org/]]

* DONE add find-by context
  :PROPERTIES:
  :suggested-by: vylu
  :END:

Hi Eric,
As I understand, we are talking here about rinari-find-by-context
function, which is used to open file by current context. For example
if I am in <%= render :partial => "something_anything" %>, it should
open _something_anything.html.erb partial. This functionality is still
left in rhtml mode (although it is broken there), but I think it
should be moved to rinari again.

Follow any of the following within two lines (up or down)
- stylesheet
- javascript
- link_to
- link_to_remote
- submit_tag
- render

* DONE improvements to finders
  :PROPERTIES:
  :suggested-by: vylu
  :END:

** DONE Stop suggesting files upon first found

#+begin_comment vylu

I see one problem probably related to latest rinari-find changes... If
specific version of a find DO NOT fail (for example, I try to open
controller from model and controller is found), requested file is
opened, but find-file promt is still left -- I have to press C-g to
close file-find or choose another file. I think, find-file promt
should be shown only if requested file is not found.

#+end_comment

** default to a more general find upon failure
When finding a view from a controller if there is no method name, then
select from the related directory.

more generally...

if rinari-find-* is called with a prefix argument then just run the
default find at the end

or even better...

*implemented* if no file is found with a find pair, then keep moving.

** TODO complex controllers
   :PROPERTIES:
   :suggested-by: rejeep
   :END:

> But say I have this controller and action:
> class Reports::Nix::UsersController < ReportsController
>   def index
>     # I am here
>   end
> end
>
> And now I do the same thing, trying to find the view:
> M-x rinari-find-view
>
> But no view is found in this case. The corresponding view in this case
> should have been app/views/reports/nix/users/index.html.erb.
>

Hi,

I've never encountered that situation before.  It sounds like when going
from a controller to a view splitting the controller name on :: and then
joining the parts as directories would handle this situation.  And then
vise-versa going from a view to a controller.  I'll try to implement
this and let you know when it's working.

Thanks for the suggestion -- Eric

* DONE upodate documentation for instillation
  :PROPERTIES:
  :suggested-by: Antti
  :END:
need to add =git submodule init; git submodule update= to clone the
submodules as well

* MAYBE windows support
  :PROPERTIES:
  :tested-by: Darren Syzling
  :END:
jumping between views, and launching servers and consoles etc... all
seem to work, the only hurdle now is a rinari-rake issue

  rinari-rake however fails with: apply: Spawning child process:
  invalid argument. Haven't had the chance to look into and debug this
  further.  I'm currently using cygwin as my shell within emacs so not
  sure if this causing some issues.

#+BEGIN_EXAMPLE
Backtrace for rinari-rake error (Spawning child process:exec format error) is:

  start-process("rake" #<buffer *rake*> "rake" "package")
  apply(start-process "rake" #<buffer *rake*> "rake" "package")
  comint-exec-1("rake" #<buffer *rake*> "rake" ("package"))
  comint-exec(#<buffer *rake*> "rake" "rake" nil ("package"))
  make-comint-in-buffer("rake" nil "rake" nil "package")
  apply(make-comint-in-buffer "rake" nil "rake" nil "package")
  make-comint("rake" "rake" nil "package")
  apply(make-comint "rake" "rake" nil "package")
  (let* ((buffer ...) (proc ...)) (save-excursion (set-buffer buffer)
(set-process-sentinel proc ...) (set-process-filter proc ...) (set ...
ruby-compilation-error-regexp-alist) (set ... ...)
(compilation-minor-mode t) (ruby-compilation-minor-mode t)))
  (if (comint-check-proc comp-buffer-name) nil (let* (... ...)
(save-excursion ... ... ... ... ... ... ...)))
  (unless (comint-check-proc comp-buffer-name) (let* (... ...)
(save-excursion ... ... ... ... ... ... ...)))
  (let ((comp-buffer-name ...)) (unless (comint-check-proc
comp-buffer-name) (let* ... ...)) comp-buffer-name)
  ruby-do-run-w/compilation("rake" ("rake" "package"))
  (pop-to-buffer (ruby-do-run-w/compilation "rake" (cons "rake" ...)))
  (let* ((task ...) (rake-args ...)) (pop-to-buffer
(ruby-do-run-w/compilation "rake" ...)))
  ad-Orig-ruby-rake-w/compilation(nil nil)
  (setq ad-return-value (ad-Orig-ruby-rake-w/compilation edit task))
  (let ((default-directory ...)) (setq ad-return-value
(ad-Orig-ruby-rake-w/compilation edit task)) (rinari-launch))
  (let (ad-return-value) (let (...) (setq ad-return-value ...)
(rinari-launch)) ad-return-value)
  ruby-rake-w/compilation(nil nil)
  rinari-rake(nil)
  call-interactively(rinari-rake)
#+END_EXAMPLE

* MAYBE inf-ruby missing errors

found an example of an error that the
`inferior-ruby-error-regexp-alist' in inf-ruby.el doesn't match

this was the error line

:SNIP:
SyntaxError: /home/eschulte/united/org/work/arf/arf/lib/cluster.rb:35: syntax error, unexpec
:END:

* MAYBE ruby-compilation missing errors
  
found a line which didn't match the error-regexp-alist

:SNIP:
test_create(QueriesControllerTest) [/home/eschulte/work/arf/arf/test/functional/queries_controller_test.rb:60]:
:END:

and another

:SNIP:
    /var/lib/gems/1.8/gems/mongrel-1.1.3/bin/mongrel_rails:281
:END:

* DONE rinari movement [5/5]

From anywhere, get to the related model/controller/test/view

where can you be?
- view
- controller
- model
- unit-test
- functional-test
- other

pending...
- [X] still need to work on movement from OTHER places
- [X] look at a better keybinding setup
- [X] rails2 style view names
- [X] code refactor:
- [X] regexps in method names

** code refactor

actually I'm thinking that when I have time it would be worthwhile to
remove toggle, and handle all of the movement through a new
independent toggle-like tool say "toggle-hassle" which uses an alist
like toggle, but instead of elements like

Done see http://github.com/eschulte/jump.el/tree/master

* DONE add yaml,css,javascript,ruby pointers to the documentation

* DONE write up some info documentation for Rinari

- [X] making progress in [[file:doc/rinari.texi]]

- [X] might want to add information for installing the documentation

- [X] also post it up to rinari.ruby-forge.org

* DONE define macro for ido completion
since we often want to check the availability of ido for completing
reads it would probably be worth writing a macro to do this.

also, it may be worthwhile breaking utility functions out into a
separate file...

* DONE Make rinari a minor mode that doesn't activate for regular ruby-mode.
  :PROPERTIES:
  :assignee: eschulte
  :END:

look at [[info:elisp:Minor%20Mode%20Conventions][info:elisp:Minor Mode Conventions]]

** DONE define as a minor mode
** DONE when to enter rinari-minor-mode
this could be done through a hook that runs every time a .rb or .rhtml
file is loaded to check if it is inside of a rails project (using the
`rails-root' function or a .emacs-project file) and then to load
rinari-minor-mode appropriately

done with hooks on find-file and dired-mode

** DONE add key-bindings for rinari minor mode
should follow emacs key bindings

see [[info:elisp:Key%20Binding%20Conventions][info:elisp:Key Binding Conventions]]

:SNIP:
   * Sequences consisting of `C-c' followed by a control character or a
     digit are reserved for major modes.

   * Sequences consisting of `C-c' followed by `{', `}', `<', `>', `:'
     or `;' are also reserved for major modes.

   * Sequences consisting of `C-c' followed by any other punctuation
     character are allocated for minor modes.  Using them in a major
     mode is not absolutely prohibited, but if you do that, the major
     mode binding may be shadowed from time to time by minor modes.
:END:

given the above, it seems like an ergonomic option would be to start
all our rinari key-bindings with \C-c-' with \C-c-'-' running
something similar to \C-c\C-c in org-mode

* DONE change toggle for incremental search

so something like search_incrementally will match search if
incrementally isn't available

also, push the "def "s into the variable in toggle.el to try to keep
it language-agnostic
* DONE project-wide search and search/replace

I know you can do this with rgrep, but It would be nice to easily call
it on the whole project (yml/rhtml/rb files), and it would help guide
emacs noobs to the powers of rgrep

Danger here is cluttering up Rinari with thin wrappers around existing
emacs tools...

rinari-rgrep

* DONE errors / tests
something to speed up the cycle of

- code
- test
- review errors

this should probably cash out into two different functions...

** DONE rails-test-method

*** DONE enhance toggle so that it can match up function names as well
see [[file:toggle.el::if%20string%20match%20new%20name][methods in toggle-buffer]]

*** DONE rinari-test-function will now try to toggle if needed

see [[file:rinari.el::defun%20rinari%20test%20function][rinari-test-function]]

*** CANCELLED use the same mechanism as find view
maybe instead of using the current buffer-file change
[[file:rinari.el::defun%20rails%20name%20components%20name][rails-name-components-name]] so it returns the last component as well,
then change the controller usage of this function to drop the last
list element.  That way this can be used to get model names when
calling which-function from models/model.rb files.

Also, if you don't have a test written for this particular method this
could insert the def...end for the test into the appropriate file, and
drop the point there.  Nothing wrong with a little bit of soft
paternalism.
* DONE add a function for running mysql
this would use the information in /conf/databases.yaml to log into the
application's database using sql-* functions from sql.el as
appropriate

* DONE make rails-find-action follow forms and link_to[_remote]

see [[file:rinari.el::defun%20rinari%20alist%20from%20view][rinari-alist-from-view]]

* DONE rinari-rake uses ruby-compilation
* DONE remove rails-script add ruby-compilation

it turns out most of the bulk of rails-script.el was supporting
running ruby processes with output going to compilation buffers.  So
rails-script.el is now removed (with two short functions in rinari.el
[[file:rinari.el::defun%20rinari%20console%20optional%20arg][rinari-console]] and [[file:rinari.el::defun%20rinari%20server][rinari-server]]), also ruby-compilation has been
added.

** DONE move through compilation buffers

redefine some keys buffers to make it easier to navigate.

n will go to the next line
p will go to the previous line

C-n next error line
C-p previous error line

M-n top of the next group of errors
M-p top of the previous group of errors [[file:rails-script.el::defun%20rails%20compilation%20previous%20error%20group][previous-group]]

so if you just saw an error on a web page, you can see that error in
three key commands

1) X-b *server*
p2) M-p
3) enter

*** old
a function which either...
1) goes to the error closest to point in the current compilation mode
   buffer
2) goes to the most recent error in the *server*, *console*, or *test*
   buffer

See this discussion below from the emacs-on-rails list...

email discussion:
:SNIP:
> - maybe add a command which can jump to the most recent error (either
>   in the server logs, or in the console)

>  > That's a great idea. find-file-at-point is pretty good at doing this
>  > from test failures in eshell, but you have to move your point up to the
>  > line containing the error message. Leveraging logs and console output
>  > would be a great addition as would looking for lines somewhere other
>  > than just under the point.
>
> So there are two different possible ways to approach this...
>
> 1) through a function which searches in known places (eshell, logs,
>    consoles, server buffers, in current buffer) and makes an educated
>    guess about what to present to the user, possibly allowing quick
>    switching to the other errors.

Hmm... now that I think about it more, I'm not sure it's a good idea to
make it really convenient to go around digging through server logs to
find errors. The right way to go about solving that kind of problem is
to write a test for it. If you make it easier to fix the problem without
writing a test, really what that amounts to is short-term benefit but
long-term problems since your fix isn't guaranteed to be caught by the
test suite.

So I would discourage integration with the dev mongrel logs. The bonus
side of that is if we only have to help jump to errors/failures in test
output it becomes a lot easier. There are in my mind three common ways
of running tests:

- In an emacs shell, whether that's eshell, M-x shell, or ansi-term
- With compile-mode, as per the ruby-test-file function in rinari
- In a shell outside Emacs

So if we check console output it will work in the first two cases but
not in the last. I think it's best to always check the test logs, since
problems are pretty much guaranteed to be found in there. This also
solves the integration test problem; when you're running those and you
get an exception, the only console output you see is that you expected a
200 response but you got a 500, which is totally useless. In those cases
you have to check the logs anyway.

We could probably base this functionality on compilation mode. It's
already built to go over output looking for something that looks like a
stack trace and provide jumping to the source of the error:

http://www.emacswiki.org/cgi-bin/wiki/CompilationMode

> 2) wrap console, and server processes in insertion-filters which could
>    notice whenever an error passes through them, and save a pointer to
>    the error in some global rails-errors variable which could be
>    queried by jump-to-my-last-error type function
>
> I was originally leaning towards the latter, but now that you mention
> it the former might be easier... not sure

Yeah, the second one sounds a lot harder.
:END:

* DONE launch web-browser to current view
  add a quick command to view the current page in a web-browser

* DONE unified runner for rails scripts

This is less a single runner, and more a single function from which
any rails/script can be executed associating the scripts with
appropriate buffers and modes.

Currently this
- runs console from a buffer in comint-mode
- runs server dumping output to a compilation-minor-mode buffer
- provides completion for generate/destroy
- runs all other scripts using shell-command-to-string dumping the
  output to a message

** DONE if web-server doesn't start change status message
currently there is no indication that the attempt to start the server
has failed

now has a sentinel, see [[elisp:(describe-function 'set-process-sentinel)]]
* DONE revert and wrap find-file-in-project
  :PROPERTIES:
  :assignee: eschulte
  :END:

revert to the .emacs-project version of find-file-in-project so that
the actual find-file-in-project.el file remains unchanged.  Then using
defadvice wrap the ffip function to find the project base using
(rails-root) as well as the .emacs-project file.

* DONE Name functions consistently
some global prefix for all of the function names, probably either
"rails" or "rinari"

going for rails for now
- easier to remember
- easy enough to change later

switching to rinari

while I'm doing this I'm going to rename rails-scripts.rl to
rails-script.el (minor)

easy enough


 LocalWords:  elsif elseif keybinding

* CANCELLED bundles
** DONE rhtml (from original rinari)
see
- [[http://rinari.rubyforge.org/rhtml_002dMode.html#rhtml_002dMode]]
- [[http://github.com/eschulte/rhtml/tree/master]]

** rdebug
- [[http://bashdb.sourceforge.net/ruby-debug.html#SEC_Top]]
** DONE snippets
see http://github.com/eschulte/yasnippets-rails/tree/master
* CANCELLED adjust eshell/shell environment to highlight ruby errors
not sure this is possible, at least not easily