| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
| |
Evidently copy-pasted from documentation.
|
| |
|
| |
|
|
|
|
|
| |
This reverts commit a14bc50. Changed my mind; decided it's tidier this
way.
|
|
|
|
| |
Preparing for spinoff release
|
| |
|
|
|
|
|
| |
We're already very dependent on Vim >=7 for this ftplugin, so we may as
well use all its syntactic sugar.
|
|
|
|
| |
This seems obvious now.
|
| |
|
|
|
|
|
|
|
| |
Because we use our own private copies of the primary filetype plugins,
they'll get loaded in the correct order from here.
Also adjust Makefile to accommodate the extra level.
|
| |
|
| |
|
|
|
|
|
|
|
| |
Per the comment in the new file, this is to avoid loading in HTML
ftplugins as well, a curiosity of the stock ftplugin/php.vim file that's
probably a well-intentioned way of accommodating templated files with a
mix of PHP and HTML in them.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is a relatively drastic change that should have been done
progressively, but I got carried away in ripping everything out and
putting it back in again.
Reading the documentation for writing a Vim script (:help usr_41.txt), I
am convinced that all of the content that was in the vim/ftplugin
directory and some of the vim/indent directory actually belonged in
vim/after/ftplugin and vim/after/indent respectively.
This is because the section on filetypes makes the distinction between
replacing the core filetype or indent plugins and merely adding to or
editing them after the fact; from :help ftplugin:
> If you do want to use the default plugin, but overrule one of the
> settings, you can write the different setting in a script:
>
> setlocal textwidth=70
>
> Now write this in the "after" directory, so that it gets sourced after
> the distributed "vim.vim" ftplugin after-directory. For Unix this
> would be "~/.vim/after/ftplugin/vim.vim". Note that the default
> plugin will have set "b:did_ftplugin", but it is ignored here.
Therefore, I have deleted the user_indent.vim and user_ftplugin.vim
plugins and their documentation that I wrote, and their ftplugin.vim and
indent.vim shims in ~/.vim, in an attempt to make these plugins
elegantly undo-ready, and instead embraced the way the documentation and
$VIMRUNTIME structure seems to suggest.
I broke the ftplugin files up by function and put them under
subdirectories of vim/after named by filetype, as the 'runtimepath'
layout permits. In doing so, I also carefully applied the
documentation's advice:
* Short-circuiting repeated loads
* Checking for existing mappings using the <Plug> prefix approach
* Avoiding repeated function declarations overwriting each other
* Guarding against 'cpotions' mangling things (by simply
short-circuiting if 'compatible' is set).
I've made the b:undo_ftplugin and b:undo_indent commands less forgiving,
and append commands to it inline with the initial establishment of the
setup they're reversing, including checking that the b:undo_* variable
actually exists in the first place.
For the indentation scripts, however, three of the four files originally
in vim/indent actually do belong there:
1. csv.vim, because it doesn't have an indent file in the core.
2. tsv.vim, because it doesn't have an indent file in the core.
3. php.vim, because it does what ftplugins are allowed to do in
preventing the core indent rules from running at all.
The indent/vim.vim rules, however, have been moved to
after/indent/vim.vim, because the tweaks it makes for two-space
indentation are designed to supplement the core indent rules, not
replace them.
Finally, I've adjusted Makefile targets accordingly for the above, given
the vim/ftplugin directory is now empty and there are three new
directories in vim/after to install. We wrap these under a single
`install-vim-after` parent target for convenience. The
`install-vim-after-ftplugin` target accommodates the additional level of
filetype directories beneath it.
|
|
|
|
| |
This was causing nasty errors whenever I started editing a Perl file.
|
|
|
|
| |
We only want to remove the normal mode mapping, since that's all we set.
|
|
|
|
|
| |
I didn't realise that a null command at the front of .e.g '|cmd|cmd2'
printed the current line! Removed that.
|
|
|
|
| |
Unload all maps too, with silent! in case they don't exist.
|
|
|
|
| |
Just for legibility.
|
|
|
|
|
| |
Just to reduce the chance of colliding with existing buffer variable
names.
|
|
|
|
|
|
| |
syntax/sh.vim only uses the existence of these variables for its checks
and as far as I can see never their actual values, so let's not overdo
things.
|
| |
|
| |
|
|
|
|
| |
This was likely a copy-paste error.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Rather than setting g:is_posix and working around core syntax/sh.vim's
ideas about Korn and POSIX shells, forego sh.vim's efforts to guess what
shell the system /bin/sh is entirely. It's irrelevant to me anyway,
since I'll often be writing shell scripts to run on an entirely
different system.
Instead, if we have a #!/bin/sh shebang reflected in the b:is_sh
variable set by core filetype.vim, and we don't have any other
buffer-level indication of what shell this is, assume it's POSIX,
because I very rarely write Bourne.
Then, after the syntax file is loaded, clear away all but one of the
resulting b:is_* variables.
I have a feeling this is going to end with me re-implementing this
syntax file, possibly as separate sh.vim, bash.vim, and ksh.vim files.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This reverts commit 09b83b6 and replaces it with a working version.
Because of the order in which the autocmd hooks run, the attempted
method of adding unloading instructions for my custom ftplugin and
indent rules to the b:undo_ftplugin and b:undo_indent doesn't actually
work.
This is because the custom rules for both groups from ~/.vim are sourced
*first*, before their core versions, so the changes the custom rules
made to b:undo_ftplugin and b:undo_indent are simply clobbered by the
core version when it loads itself.
Therefore we need to arrange for two things:
1. A custom variable needs to be checked and executed when the filetype
changes to revert the changes for the custom ftplugin or indent
rules.
2. That execution needs to take place *first* when the filetype
changes.
I wrote two simple plugins with very similar code that are designed to
run as a user's custom ftplugin.vim and indent.vim implementations,
running before their brethren in the Vim core, and setting up an autocmd
hook to :execute b:undo_user_ftplugin and b:undo_user_indent plugin
respectively.
This seemed to work well, so I've implemented it. It involves adding a
shim to ~/.vim/indent.vim and ~/.vim/ftplugin.vim to "preload" the
plugin when the `filetype indent plugin on` call is made. I've added
that to the relevant Makefile targets.
|
|
|
|
|
|
|
|
|
|
|
| |
Setting or adding to b:undo_indent and b:undo_ftplugin variables, which
I only learned about just now, allows me to avoid the _GLOBAL.vim hack
and remove some files from both vim/indent/ and vim/ftplugin/.
These variables aren't subjected to :execute automatically in anything
older than Vim 7.0, but I don't think that's too much of a concern as
the only real reason they're needed are for changing filetypes in the
same buffer, which doesn't happen that often anyway.
|
|
|
|
| |
Looks like this was mistakenly omitted in commit 09f8635.
|
|
|
|
|
| |
Put the entire command line for the determined check and lint into the
variable, so it can just be directly executed.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Including renaming big_file.vim and accompanying functions yet again, to
big_file_options.vim.
Trying to keep complex autocmd and mapping definitions on long lines
broken up semantically; definition and options on one line, patterns or
mapping key on the next, and the command to run on the last.
Also trying to make sure that <silent>, <buffer>, and <unique> are
applied in the correct places, and that all mapping commands are using
the :<C-U> idiom for the command prefix.
|
|
|
|
| |
Runs `vint -s`; the -s includes stylistic suggestions.
|
|
|
|
|
| |
The commands to use in this case are dependent on the particular shell
being used.
|
|
|
|
|
|
|
|
|
|
|
| |
This mapping mirrors the one for Perl that passes the content of the
buffer through a program to tidy it (i.e. not merely check but actively
change it).
The tidy(1) option chosen here, -quiet, is the bare minimum to make this
invocation useful. We would never want the boilerplate it otherwise
emits to be in the buffer after a call. Everything else should be
applied in a configuration file, which I'll do in a separate feature.
|
| |
|
|
|
|
|
|
| |
That is, apply <buffer> and <silent> to each of them, to make them only
apply to the current buffer and to prevent them from echoing the command
they're running.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
This is a much better method of calling external programs on the
buffer's contents, not just because it avoids the mess of :execute
evaluation but also because it doesn't require that there actually be a
filename for the current buffer.
This drastically simplifies the HTML tidy(1) call in particular.
|
|
|
|
| |
We should probably avoid this sort of abbreviation in scripts.
|
|
|
|
|
| |
It doesn't seem to be in very old Vims; worth testing for to avoid
errors if I try to use the function.
|
|
|
|
|
|
|
|
| |
I thought text objects were introduced to Vim a lot later than they
actually were; this works fine even in Vim 6, so I'll leave it as it's
nicer.
This reverts commit ffb5cbc7c681e2fdcb780dbdc51cf3458a937791.
|
| |
|
|
|
|
|
|
|
|
|
| |
Tidy up the 'normal!' commands and comment them in appropriate groups.
Take advantage of the natural command-termination at the end of a
'normal!' string to end insert mode.
It would be better to do all of this with pure VimL functions, but I
don't know how yet.
|
|
|
|
| |
Saves some backslashing, just like in shell and Perl!
|
|
|
|
|
| |
Move the logic into a script function. Use single quotes for the
strings, too, since we don't need interpolation.
|
|
|
|
|
|
|
| |
From ":help <LocalLeader>":
> In a global plugin <Leader> should be used and in a filetype plugin
> <LocalLeader>. "mapleader" and "maplocalleader" can be equal.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The Google VimScript Style Guide says
<https://google.github.io/styleguide/vimscriptguide.xml#Naming>:
>In general, use plugin-names-like-this, FunctionNamesLikeThis,
>CommandNamesLikeThis, augroup_names_like_this,
>variable_names_like_this.
Adjusted variable, function, and `augroup` names accordingly, including
setting script scope for some of the functions and their calls (`s:` and
`<SID>` prefixes).
Initially I tried using `prefix#`, but it turns out that this is a
namespacing contention for publically callable functions like
`pathogen#infect`, and none of these functions need to be publically
callable.
|
|\
| |
| |
| |
| |
| | |
The Vim configuration, excluding the submodule plugin bundles, now
passes a strict run of the vim-vint tool. There's also now a `lint-vim`
target in the Makefile.
|
|/
|
|
|
| |
This was mistakenly moved along with some indentation settings in
9858af6.
|