docs: Improve plugin documentation #3240

glupi-borna · 2024-04-08T18:57:25Z

Added direct links to relevant external documentation pages, rewrote some sections, documented missing functions/callbacks, reformatted everything to fit in 80 columns or less. Attempted to optimize for legibility within micro, without sacrificing what the rendered markdown looks like.

This should address some of the stuff that came to light in #3231.

I think the layout of the rendered markdown of the package/callback docs could be greatly improved, but it is hard to do so without hurting the readability within micro.

dmaluka · 2024-04-09T23:56:32Z

runtime/help/plugins.md

-from the go plugin, which has the following file structure:
+provides additional information, such as the name of the plugin, the
+plugin's website, dependencies, etc.
+[Here is an example `repo.json` file][repo.json] from the go plugin, which has


It took me some time to figure out that this [repo.json] is a reference to the "footnote" below.

As you noted yourself, we need to care about readability of this help inside micro, i.e. as plain text, not just as rendered markdown. So I'm not sure this change is a good idea.

Yeah, I thought it might not be totally clear what's going on there, but I decided to go with it anyway because I thought the links break the flow and layout of the text a bit.

But that's my bad, I'll get rid of this type of link. Would you prefer I keep the links to the golang documentation in some form (as literal footnotes maybe?), or should I completely get rid of them?

Also, thanks so much for taking the time to review this.

Would you prefer I keep the links to the golang documentation in some form (as literal footnotes maybe?), or should I completely get rid of them?

I'm not against keeping them in some form, I do agree that providing more granular links than just the current Please go to the internal documentation at https://pkg.go.dev/github.com/zyedidia/micro/v2/internal to see the full list of available methods. should make the documentation more convenient to use.

dmaluka · 2024-04-09T23:59:49Z

runtime/help/plugins.md

-syntax header files. No directory structure is enforced but keeping
-runtime files in their own directories is good practice.
+cases, a plugin may also want to add colorschemes or syntax files. It
+is unlikely for a plugin to need to add plugin or syntax header files at


I think this sentence meant to say that a plugin is unlikely to load syntax header files at all. Actually, a plugin is not even able to load them, since RTSyntaxHeader is not exported to Lua, unlike other runtime file types. So I think we should just remove "or syntax header files" here.

Nice catch! I didn't even RTSyntaxHeader is not exported.

dmaluka · 2024-04-10T00:07:48Z

runtime/help/plugins.md

-   initialized.
+* `deinit()`:
+   cleanup function called when your plugin is unloaded, either manually by the
+   user, or by running the `reload` command.


I'm ok with changing the style (function description beginning not at the same line but at the next line), but please use consistently the same indentation everywhere.

Also could you move these style changes (that are moving the text without changing its contents) to a separate commit, to separate them from the actual text changes? It's really hard to review when the two are mixed.

Yep, you're right. I'll amend this PR to remove the changes, and then I'll make another one with just the layout/styling changes at some point.

Just in case, I didn't mean necessarily a separate PR. It can be done in one PR, just in separate commits.

With 2b384de it looks good now, but could you squash 2b384de into the previous commit, to make it easy to review? (Actually I already ended up doing that myself locally and reviewing via git difftool. But what if others want to review it too.)

Good call! I was going to squash it but I guess I got distracted and forgot.

Sorry for the delay, didn't realize you already squashed it. The damn github didn't send any notification.

No worries at all! :)

dmaluka · 2024-04-10T00:10:53Z

runtime/help/plugins.md


-* `postinit()`: initialization function called after `init()`.
+* `onSetActive(`[bufpane][BufPane]`)`:


Again, a user reading help plugins in micro (in plain text) will have no idea what the heck is this [bufpane][BufPane].

Having the links in the rendered markdown is extremely useful though so maybe the tradeoff is worth it?

I think it is not a tradeoff but the opposite of it, i.e. favoring one use case at the expense of the other one.

A tradeoff would be something like: do add these links, but only as "footnotes". Perhaps also with a sentence like See these links for documentation on these data structures:.

Right, the primary use case for the docs is to be viewed from within micro, so we should optimize for that and strive to remove as much visual noise as possible. Maybe the links could be kept somewhere close by, but not within the callback signatures themselves.

dmaluka · 2024-04-10T00:12:53Z

runtime/help/plugins.md

       stdout from the command to the returned string.

    - `JobStart(cmd string, onStdout, onStderr,
                onExit func(string, []interface{}), userargs ...interface{})
-                *exec.Cmd`:
+                *Cmd`:


After adding links, it seemed to me that the exec. (and possibly buffer. and display.) were just unnecessary noise, but that might just be my perspective. In fact, I think that just dumping go function signatures directly into the plugin docs is not that good of an idea, seeing as plugins are written in Lua. It expects that plugin authors understand at least some subset of go, which is likely not true in a lot of cases.

That said, I decided not to change the layout of all the function signatures because it seemed too big of a change for what I'm trying to achieve here (in retrospect, I should have left the layout alone too and just focused on the content of the docs). I did attempt to do it (for a little while at least), just to explore the possibility space a bit, but I didn't come up with a strictly better layout (in regards to being readable in both micro and in a markdown viewer).

In fact, I think that just dumping go function signatures directly into the plugin docs is not that good of an idea, seeing as plugins are written in Lua.

But this is not just about docs, it is a deeper "issue". The fact is that although plugin authors write their plugins in Lua, for the most part they use APIs that are defined in Go, not in Lua. Just because that's how Micro's plugin system is designed, for good or bad. So the docs should not pretend that it's not like that, I think.

It expects that plugin authors understand at least some subset of go, which is likely not true in a lot of cases.

But the reality is that they do need to understand at least some subset of go (see above). At least to understand signatures of Buffer or BufPane methods, and so on.

dmaluka · 2024-04-10T00:15:45Z

runtime/help/plugins.md

+There are 7 default plugins that come pre-installed with micro. These are
+
+* `autoclose`:
+   automatically closes brackets, quotes, etc...


I'm not sure that moving these to separate lines here really makes it more readable.

I think I moved them because for consistency with the function descriptions. Now that I look at it, it really does not improve anything.

dmaluka · 2024-05-09T17:16:00Z

runtime/help/plugins.md

@@ -51,6 +50,9 @@ which micro defines:

 * `postinit()`: initialization function called after `init()`.

+* `deinit()`: cleanup function called when your plugin is unloaded,
+   either manually by the user, or by running the `reload` command.


It's not clear what does it mean "unloading manually" (I ended up looking into the code to recall what it means).

So, it means "disabling plugin via the set command", right?

BTW, the reload command is also run by the user, and also manually, right? :)

BTW preinit(), init() and postinit() are also called when running the reload command. It doesn't mean that we need to mention it in their above descriptions as well, but we should probably better document the reload command itself, in help commands. Currently it is:

* `reload`: reloads all runtime files.

It's likely not clear to the user what is "runtime" files. So maybe let's make it clear:

* `reload`: reloads all runtime files (settings, keybindings, syntax files, colorschemes, plugins).

And then maybe also document its behavior wrt callbacks, something like:

* `reload`: reloads all runtime files (settings, keybindings, syntax files, colorschemes, plugins). For plugins, it properly unloads already loaded plugins (by running their `deinit()` callbacksif there are any) and then re-initializes reloaded plugins (by running their `preinit()`, `init()` and `postinit()` callbacks if there are any).

Please feel free to suggest a better wording.

BTW, do we document anywhere that the set command can be used to enable/disable plugins, not just options? Looks like we don't. (I thought we did...) So it makes sense to document it (both in help plugins and in the description of the set command in help commands)?

So, it means "disabling plugin via the set command", right?

Correct, and yeah, good point about documenting set and reload better.

I mean help options does say that set can be used to enable/disable plugins, but it's only obvious if you read it immediately after reading about set in help commands:

help commands

[...]

set 'option' 'value': sets the option to value. See the options help topic for a list of options you can set. This will modify your settings.json with the new value.

[...]

help options

[...]
Plugin options: all plugins come with a special option to enable or disable
them. The option is a boolean with the same name as the plugin itself.
[...]

However, this is super unclear and roundabout, and might confuse new users. It would definitely be much better to have the behavior explicitly documented. I'll give it a shot.

help options

[...]
Plugin options: all plugins come with a special option to enable or disable
them. The option is a boolean with the same name as the plugin itself.
[...]

Ah, indeed. So it's actually documented, so it's not so bad. So maybe it's not necessary to lengthen the description of the set command with an explicit mentioning of plugin options. Still, I guess we should add mentioning them at least to help plugins, since it is, well, about plugins. (I.e. help plugins is probably the place where users would look for info how to disable a plugin and so on.)

I had some trouble figuring out how to naturally fit this into plugins.md -- I ended up just adding a note to the top of the file, clarifying that plugins.md describes how to create plugins, and then directed the user to either help options or help commands, depending on what they're looking for. Hopefully that is good enough, unless you have a better idea?

LGTM, thanks.

dmaluka · 2024-05-09T17:22:14Z

runtime/help/plugins.md

@@ -51,6 +50,9 @@ which micro defines:

 * `postinit()`: initialization function called after `init()`.


While we're at it, shouldn't we improve this description: clarify that it is called after the init() callbacks of all other plugins have been called? Otherwise it's not clear what's the point of this postinit() callback.

Absolutely, makes sense.

glupi-borna · 2024-06-08T14:18:20Z

@dmaluka sorry this took so long! I was very busy these past few weeks. I have implemented the changes we have talked about, and also documented the reopen command. Check it out when you get the chance :)

dmaluka · 2024-06-10T15:37:46Z

runtime/help/commands.md


 * `cd 'path'`: Change the working directory to the given `path`.

 * `pwd`: Print the current working directory.

 * `open 'filename'`: Open a file in the current buffer.

+* `reopen`: Reopens the current buffer from disk.


nit: "Reopens the current file from disk" would be more accurate?

Yep, both this and your other nit make sense. I was actually going to write it the way you did, but decided to be conservative and just slightly adapt the doc comment for ReopenCmd:

// ReopenCmd reopens the buffer (reload from disk) func (h *BufPane) ReopenCmd(args []string)

Anyway, I went ahead and changed both to the nicer version.

dmaluka · 2024-06-10T15:49:07Z

runtime/help/plugins.md

    - `BTRaw`: raw buffer type.
    - `BTInfo`: info buffer type.

    - `NewBuffer(text, path string) *Buffer`: creates a new buffer with the
       given text at a certain path.

    - `NewBufferFromFile(path string) (*Buffer, error)`: creates a new
-       buffer by reading from disk at the given path.
+       buffer by reading from disk at the given path. Returns an error if


nit (cont.): "by reading a file from disk at the given path"?

`commands.md` - documented the `reopen` command - improved the documentation of the `reload` command `plugins.md` - added direct links to relevant external documentation pages - rewrote some sections - documented missing functions/callbacks - added a note about installing/managing plugins

runtime/help/plugins.md

Co-authored-by: Jöran Karl <[email protected]>

dmaluka reviewed Apr 10, 2024

View reviewed changes

glupi-borna force-pushed the docs-improvements branch from 2b384de to 7a06fdf Compare April 21, 2024 11:41

glupi-borna mentioned this pull request Apr 23, 2024

Do we need to track duplicated plugin descriptions in options.md & plugins.md? #3217

Open

dmaluka reviewed May 9, 2024

View reviewed changes

dmaluka mentioned this pull request May 19, 2024

Garbled contents when reloading file with DOS line endings #3303

Open

niten94 mentioned this pull request Jun 4, 2024

Running shell script in statusformatl #1860

Open

glupi-borna force-pushed the docs-improvements branch from 7a06fdf to 239b78b Compare June 8, 2024 14:15

dmaluka reviewed Jun 10, 2024

View reviewed changes

glupi-borna force-pushed the docs-improvements branch from 239b78b to 3727fad Compare June 10, 2024 18:35

dmaluka approved these changes Jun 10, 2024

View reviewed changes

dmaluka mentioned this pull request Jun 17, 2024

Rework filetype change, reload command and autosave #3343

Merged

JoeKar reviewed Jun 18, 2024

View reviewed changes

runtime/help/plugins.md Outdated Show resolved Hide resolved

Omit number of default plugins

c46e612

Co-authored-by: Jöran Karl <[email protected]>

JoeKar merged commit a3e25e3 into zyedidia:master Jun 19, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

docs: Improve plugin documentation #3240

docs: Improve plugin documentation #3240

glupi-borna commented Apr 8, 2024

dmaluka Apr 9, 2024

glupi-borna Apr 10, 2024

dmaluka Apr 10, 2024

dmaluka Apr 9, 2024

glupi-borna Apr 10, 2024

dmaluka Apr 10, 2024

glupi-borna Apr 10, 2024

dmaluka Apr 10, 2024

dmaluka Apr 20, 2024

glupi-borna Apr 21, 2024

dmaluka May 9, 2024

glupi-borna May 10, 2024

dmaluka Apr 10, 2024

Andriamanitra Apr 10, 2024

dmaluka Apr 10, 2024

glupi-borna Apr 10, 2024

dmaluka Apr 10, 2024

glupi-borna Apr 10, 2024

dmaluka Apr 10, 2024

dmaluka Apr 10, 2024

glupi-borna Apr 10, 2024

dmaluka May 9, 2024

dmaluka May 9, 2024

glupi-borna May 10, 2024

dmaluka May 10, 2024

glupi-borna Jun 8, 2024

dmaluka Jun 10, 2024

dmaluka May 9, 2024

glupi-borna May 10, 2024

glupi-borna commented Jun 8, 2024

dmaluka Jun 10, 2024

glupi-borna Jun 10, 2024

dmaluka Jun 10, 2024


		* `postinit()`: initialization function called after `init()`.
		* `onSetActive(`[bufpane][BufPane]`)`:

		@@ -51,6 +50,9 @@ which micro defines:

		* `postinit()`: initialization function called after `init()`.

docs: Improve plugin documentation #3240

docs: Improve plugin documentation #3240

Conversation

glupi-borna commented Apr 8, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

glupi-borna commented Jun 8, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment