I am new to Geany plugins and GTK development in general, and I'd like to know some best practices (and how-to's) for a plugin I'm working on. It does code folding, by level--that is, rather than folding all of the code at once, just the top-level or second-level (or third-level...). (This is very useful to me, as I can open a file and get all the functions at a glance, or e.g. for a python class look at all the function defs by folding the second level, and so on.) I did so by essentially copying the folding code in Geany's fold_all function, and then adding a submenu and keybindings to call the function.
I think this may be useful to others and I'd like to submit it to geany-plugins, but I have a couple questions first:
1) The submenu is in the tools menu, but I feel it would be better in the Document menu underneath the existing folding functions. Is this a good idea, and if so, how to I get at the document menu (since I can't say geany->main_widgets->tools_menu)?
2) I didn't set the keybindings since it's not recommended as it may interfere with existing ones. Is there a way to set them if the given key is not in use, or is that a bad idea?
3) In the actual folding code I used sci_get_fold_level (and a couple other scintilla wrappers) which aren't in the plugin API so I gather my plugin may break at some point. There doesn't seem to be a way around this for what I want to do. Is this ok?
4) Is there another plugin out there that does this? I couldn't find one.
5) Would geany-plugins accept another plugin?
Thanks, Peter O'Malley
4. The symbols tab on the side panel shows an overview of functions. You can change the tags to be listed in order of occurance instead of alphabetical in a config file. 5. I am guessing geany-plugins welcomes more plugins, so long as the dev agrees to maintain their plugin.
Thanks,
Steve
On Wed, 15 Jan 2014 19:24:31 -0800 (PST) Steven Blatnick steve8track@yahoo.com wrote:
5. I am guessing geany-plugins welcomes more plugins, so long as the dev agrees to maintain their plugin.
Yes. New plugins are welcome as long as they have a minimum in quality and will be maintained in future.
Cheers, Frank
On 20 January 2014 01:24, Frank Lanitz frank@frank.uvena.de wrote:
On Wed, 15 Jan 2014 19:24:31 -0800 (PST) Steven Blatnick steve8track@yahoo.com wrote:
- I am guessing geany-plugins welcomes more plugins, so long as the
dev agrees to maintain their plugin.
Yes. New plugins are welcome as long as they have a minimum in quality and will be maintained in future.
That is *at least* a minimum quality :)
Cheers Lex
Cheers, Frank
Devel mailing list Devel@lists.geany.org https://lists.geany.org/cgi-bin/mailman/listinfo/devel
Am 19.01.2014 22:08, schrieb Lex Trotman:
On 20 January 2014 01:24, Frank Lanitz <frank@frank.uvena.de mailto:frank@frank.uvena.de> wrote:
On Wed, 15 Jan 2014 19:24:31 -0800 (PST) Steven Blatnick <steve8track@yahoo.com <mailto:steve8track@yahoo.com>> wrote: > 5. I am guessing geany-plugins welcomes more plugins, so long as the > dev agrees to maintain their plugin. Yes. New plugins are welcome as long as they have a minimum in quality and will be maintained in future.That is *at least* a minimum quality :)
Yepp. What Lex says ;)
Cheers, Frank
OK, thanks everyone. I will follow your advice :-). (Though I am a bit skeptical about using the scintilla API directly, but I guess that's less likely to change.)
I will hopefully soon have something completed to work with, but at the moment I'm having trouble getting my plugin to compile as part of geany-plugins; for some reason GETTEXT_PACKAGE is undeclared even though LOCALEDIR is fine. Once I have enough time to sort that out I'll create a pull request so y'all can take a look and see if you're interested in it.
Thanks, Peter
On Fri, Jan 24, 2014 at 4:20 AM, Frank Lanitz frank@frank.uvena.de wrote:
Am 19.01.2014 22:08, schrieb Lex Trotman:
On 20 January 2014 01:24, Frank Lanitz <frank@frank.uvena.de mailto:frank@frank.uvena.de> wrote:
On Wed, 15 Jan 2014 19:24:31 -0800 (PST) Steven Blatnick <steve8track@yahoo.com <mailto:steve8track@yahoo.com>> wrote: > 5. I am guessing geany-plugins welcomes more plugins, so long as the > dev agrees to maintain their plugin. Yes. New plugins are welcome as long as they have a minimum in quality and will be maintained in future.That is *at least* a minimum quality :)
Yepp. What Lex says ;)
Cheers, Frank
Devel mailing list Devel@lists.geany.org https://lists.geany.org/cgi-bin/mailman/listinfo/devel
Le 24/01/2014 17:56, Peter O'Malley a écrit :
[...]
I will hopefully soon have something completed to work with, but at the moment I'm having trouble getting my plugin to compile as part of geany-plugins; for some reason GETTEXT_PACKAGE is undeclared even though LOCALEDIR is fine. Once I have enough time to sort that out I'll create a pull request so y'all can take a look and see if you're interested in it.
Do you include config.h?
Regards, Colomban
Turns out I'm pretty rusty working in C... I knew it would be something silly. Thanks!
I have another question: the code folding works fine, but I'm having an issue with how the fold levels are counted in python. The top block is still level 1, but the next block is level 5, I gather because it's indented four spaces. I'd prefer if it's counted as level two. I realize that due to the rules python has about indenting it makes sense for the parser to work this way. However, is there any obvious way I can change it? Or would I have to either go deep into Scintilla to "fix" something that isn't broken, or try to work around it at the top level? I'm basically wondering if I'm missing anything obvious.
Thanks, Peter
On Fri, Jan 24, 2014 at 9:12 AM, Colomban Wendling < lists.ban@herbesfolles.org> wrote:
Le 24/01/2014 17:56, Peter O'Malley a écrit :
[...]
I will hopefully soon have something completed to work with, but at the moment I'm having trouble getting my plugin to compile as part of geany-plugins; for some reason GETTEXT_PACKAGE is undeclared even though LOCALEDIR is fine. Once I have enough time to sort that out I'll create a pull request so y'all can take a look and see if you're interested in it.
Do you include config.h?
Regards, Colomban _______________________________________________ Devel mailing list Devel@lists.geany.org https://lists.geany.org/cgi-bin/mailman/listinfo/devel
On 25 January 2014 07:20, Peter O'Malley ooomalley@gmail.com wrote:
Turns out I'm pretty rusty working in C... I knew it would be something silly. Thanks!
I have another question: the code folding works fine, but I'm having an issue with how the fold levels are counted in python. The top block is still level 1, but the next block is level 5, I gather because it's indented four spaces. I'd prefer if it's counted as level two. I realize that due to the rules python has about indenting it makes sense for the parser to work this way. However, is there any obvious way I can change it? Or would I have to either go deep into Scintilla to "fix" something that isn't broken, or try to work around it at the top level? I'm basically wondering if I'm missing anything obvious.
Deep in LexPython.cxx :)
But consider:
1) The level number the folder produces is for folding purposes, not Python syntax, so for eg successive lines of triple quoted strings have the level increased by one to cause the string to be foldable, even though the actual source is not indented further.
2) The folder only works for PEP8 styled files, some legal but non-PEP-8 implicit or explicit line continuation situations are folded wrongly
3) An increase in the level number is the fold point, not what the number's value is. Or you could read the SC_FOLDLEVELHEADERFLAG flag which erm ... flags the start of a fold.
4) The folder would have to always count increases/decreases from the start of the file to actually get a level that counts correctly or save state on the file. At the moment it starts from near the start of the visible range most of the time to make it faster, and saves no state so it only knows the actual indentation, not the level number. You will have to keep count yourself as you go through the file clearing/setting the fold points.
Cheers Lex
Thanks, Peter
On Fri, Jan 24, 2014 at 9:12 AM, Colomban Wendling < lists.ban@herbesfolles.org> wrote:
Le 24/01/2014 17:56, Peter O'Malley a écrit :
[...]
I will hopefully soon have something completed to work with, but at the moment I'm having trouble getting my plugin to compile as part of geany-plugins; for some reason GETTEXT_PACKAGE is undeclared even though LOCALEDIR is fine. Once I have enough time to sort that out I'll create a pull request so y'all can take a look and see if you're interested in it.
Do you include config.h?
Regards, Colomban _______________________________________________ Devel mailing list Devel@lists.geany.org https://lists.geany.org/cgi-bin/mailman/listinfo/devel
Devel mailing list Devel@lists.geany.org https://lists.geany.org/cgi-bin/mailman/listinfo/devel
On Fri, Jan 24, 2014 at 2:29 PM, Lex Trotman elextr@gmail.com wrote:
... 4) The folder would have to always count increases/decreases from the start of the file to actually get a level that counts correctly or save state on the file. At the moment it starts from near the start of the visible range most of the time to make it faster, and saves no state so it only knows the actual indentation, not the level number. You will have to keep count yourself as you go through the file clearing/setting the fold points.
Cheers Lex
Let's say that I'm only interested in python code that's (close to) PEP 8 and ignore stuff like triple-quoted strings for the moment. "Keeping count myself" throughout the entire file is basically the only idea I had come up with. I was thinking of something fairly simple like this: * User requested to fold level 2 * Check document for HEADERFLAG's at fold level 1; found some; increment counter * Check level 2... none exist, level 3... none exist... (etc) found some at level 5 * Counter now at 2, so fold all level 5
Obviously this wouldn't scale too well for large files. But in my (limited) experience python files don't grow very large... and I personally don't like large source files anyway ;-). (And I suppose this functionality could be turned on/off with a setting, too.)
How does this sound?
And thanks for the pointers, too!
Best, Peter
On 26 January 2014 02:53, Peter O'Malley ooomalley@gmail.com wrote:
On Fri, Jan 24, 2014 at 2:29 PM, Lex Trotman elextr@gmail.com wrote:
... 4) The folder would have to always count increases/decreases from the
start
of the file to actually get a level that counts correctly or save state
on
the file. At the moment it starts from near the start of the visible
range
most of the time to make it faster, and saves no state so it only knows
the
actual indentation, not the level number. You will have to keep count yourself as you go through the file clearing/setting the fold points.
Cheers Lex
Let's say that I'm only interested in python code that's (close to) PEP 8 and ignore stuff like triple-quoted strings for the moment.
Ok, just thought docstrings might be useful for you to show.
"Keeping count myself" throughout the entire file is basically the only idea I had come up with. I was thinking of something fairly simple like this:
- User requested to fold level 2
- Check document for HEADERFLAG's at fold level 1; found some; increment
counter
- Check level 2... none exist, level 3... none exist... (etc) found
some at level 5
- Counter now at 2, so fold all level 5
Obviously this wouldn't scale too well for large files. But in my (limited) experience python files don't grow very large...
wrong :)
If its put in Geany-Plugins you *will* get bug reports if its slow with big files. Remember you are maintaining it :)
If I read your above right, you are scanning the file multiple times. And what is triggering this? How often will it run and annoy your users? But if its only triggered manually I would have said its ok.
But in any case, naively I would have said only one scan is needed using the levels from the current lexer (warning, not much thought gone into this :)
level = 0 fold_level = 0 for all lines: line_level = get_fold_level_of_line() if fold_level < line_level): ++level elif fold_level > line_level: --level fold_level = line_level if line has header_flag: if level < level_to_expand_to: unfold else fold
and I personally don't like large source files anyway ;-).
Again if you are making it public via G-P it needs "a reasonable level of quality" since its not just your files any more :)
(And I suppose this functionality could be turned on/off with a setting, too.)
Well, its a plugin, so it can be disabled. Thats fine.
Cheers Lex
How does this sound?
And thanks for the pointers, too!
Best, Peter _______________________________________________ Devel mailing list Devel@lists.geany.org https://lists.geany.org/cgi-bin/mailman/listinfo/devel
Again if you are making it public via G-P it needs "a reasonable level of quality"
I appreciate the help, Lex. What I meant for the scanning was that it would happen twice, once to determine that, e.g. fold level 5 is actually fold level 2 and fold level 9 is actually fold level 3, and then again to do the actual folding. I think your outline of folding-as-you-go would probably work, too, and may, in fact, be better.
It would only scan when the user initiates the (un)fold all at level X action. Geany scans the whole file for the (un)fold all action already, and given that that doesn't take too long, I'm guessing that this wouldn't either. But I haven't tried it out yet.
Peter
On Sat, Jan 25, 2014 at 3:13 PM, Lex Trotman elextr@gmail.com wrote:
On 26 January 2014 02:53, Peter O'Malley ooomalley@gmail.com wrote:
On Fri, Jan 24, 2014 at 2:29 PM, Lex Trotman elextr@gmail.com wrote:
... 4) The folder would have to always count increases/decreases from the
start
of the file to actually get a level that counts correctly or save state
on
the file. At the moment it starts from near the start of the visible
range
most of the time to make it faster, and saves no state so it only knows
the
actual indentation, not the level number. You will have to keep count yourself as you go through the file clearing/setting the fold points.
Cheers Lex
Let's say that I'm only interested in python code that's (close to) PEP 8 and ignore stuff like triple-quoted strings for the moment.
Ok, just thought docstrings might be useful for you to show.
"Keeping count myself" throughout the entire file is basically the only idea I had come up with. I was thinking of something fairly simple like this:
- User requested to fold level 2
- Check document for HEADERFLAG's at fold level 1; found some; increment
counter
- Check level 2... none exist, level 3... none exist... (etc) found
some at level 5
- Counter now at 2, so fold all level 5
Obviously this wouldn't scale too well for large files. But in my (limited) experience python files don't grow very large...
wrong :)
If its put in Geany-Plugins you *will* get bug reports if its slow with big files. Remember you are maintaining it :)
If I read your above right, you are scanning the file multiple times. And what is triggering this? How often will it run and annoy your users? But if its only triggered manually I would have said its ok.
But in any case, naively I would have said only one scan is needed using the levels from the current lexer (warning, not much thought gone into this :)
level = 0 fold_level = 0 for all lines: line_level = get_fold_level_of_line() if fold_level < line_level): ++level elif fold_level > line_level: --level fold_level = line_level if line has header_flag: if level < level_to_expand_to: unfold else fold
and I personally don't like large source files anyway ;-).
Again if you are making it public via G-P it needs "a reasonable level of quality" since its not just your files any more :)
(And I suppose this functionality could be turned on/off with a setting, too.)
Well, its a plugin, so it can be disabled. Thats fine.
Cheers Lex
How does this sound?
And thanks for the pointers, too!
Best, Peter _______________________________________________ Devel mailing list Devel@lists.geany.org https://lists.geany.org/cgi-bin/mailman/listinfo/devel
Again if you are making it public via G-P it needs "a reasonable level of quality"
Devel mailing list Devel@lists.geany.org https://lists.geany.org/cgi-bin/mailman/listinfo/devel
On 28 January 2014 10:00, Peter O'Malley ooomalley@gmail.com wrote:
I appreciate the help, Lex. What I meant for the scanning was that it would happen twice, once to determine that, e.g. fold level 5 is actually fold level 2 and fold level 9 is actually fold level 3, and then again to do the actual folding.
Ok, but you are assuming that file level say 5 will be fold level say 3 throughout the file.
I think your outline of folding-as-you-go would probably work, too, and may, in fact, be better.
The important thing is the algorithm comparing if the line level is greater than or less than the previous line, thats what Python actually does. Note it doesn't actually care what the numbers are, just if they increase or decrease.
It would only scan when the user initiates the (un)fold all at level X action. Geany scans the whole file for the (un)fold all action already, and given that that doesn't take too long, I'm guessing that this wouldn't either. But I haven't tried it out yet.
Ok, scanning all but *very* big files will likely be less than the time to re-render the screen.
Cheers Lex
Peter
On Sat, Jan 25, 2014 at 3:13 PM, Lex Trotman elextr@gmail.com wrote:
On 26 January 2014 02:53, Peter O'Malley ooomalley@gmail.com wrote:
On Fri, Jan 24, 2014 at 2:29 PM, Lex Trotman elextr@gmail.com wrote:
... 4) The folder would have to always count increases/decreases from the
start
of the file to actually get a level that counts correctly or save
state on
the file. At the moment it starts from near the start of the visible
range
most of the time to make it faster, and saves no state so it only
knows the
actual indentation, not the level number. You will have to keep count yourself as you go through the file clearing/setting the fold points.
Cheers Lex
Let's say that I'm only interested in python code that's (close to) PEP 8 and ignore stuff like triple-quoted strings for the moment.
Ok, just thought docstrings might be useful for you to show.
"Keeping count myself" throughout the entire file is basically the only idea I had come up with. I was thinking of something fairly simple like this:
- User requested to fold level 2
- Check document for HEADERFLAG's at fold level 1; found some; increment
counter
- Check level 2... none exist, level 3... none exist... (etc) found
some at level 5
- Counter now at 2, so fold all level 5
Obviously this wouldn't scale too well for large files. But in my (limited) experience python files don't grow very large...
wrong :)
If its put in Geany-Plugins you *will* get bug reports if its slow with big files. Remember you are maintaining it :)
If I read your above right, you are scanning the file multiple times. And what is triggering this? How often will it run and annoy your users? But if its only triggered manually I would have said its ok.
But in any case, naively I would have said only one scan is needed using the levels from the current lexer (warning, not much thought gone into this :)
level = 0 fold_level = 0 for all lines: line_level = get_fold_level_of_line() if fold_level < line_level): ++level elif fold_level > line_level: --level fold_level = line_level if line has header_flag: if level < level_to_expand_to: unfold else fold
and I personally don't like large source files anyway ;-).
Again if you are making it public via G-P it needs "a reasonable level of quality" since its not just your files any more :)
(And I suppose this functionality could be turned on/off with a setting, too.)
Well, its a plugin, so it can be disabled. Thats fine.
Cheers Lex
How does this sound?
And thanks for the pointers, too!
Best, Peter _______________________________________________ Devel mailing list Devel@lists.geany.org https://lists.geany.org/cgi-bin/mailman/listinfo/devel
Again if you are making it public via G-P it needs "a reasonable level of quality"
Devel mailing list Devel@lists.geany.org https://lists.geany.org/cgi-bin/mailman/listinfo/devel
Devel mailing list Devel@lists.geany.org https://lists.geany.org/cgi-bin/mailman/listinfo/devel
[...]
- The submenu is in the tools menu, but I feel it would be better in
the Document menu underneath the existing folding functions. Is this a good idea, and if so, how to I get at the document menu (since I can't say geany->main_widgets->tools_menu)?
In general its not a good idea to hide your menu items in existing menus, users don't know where to look.
- I didn't set the keybindings since it's not recommended as it may
interfere with existing ones. Is there a way to set them if the given key is not in use, or is that a bad idea?
How do you know that the user hasn't defined this key for the *next* plugin to be loaded after yours, the loading order is not defined? No, bad idea.
- In the actual folding code I used sci_get_fold_level (and a couple
other scintilla wrappers) which aren't in the plugin API so I gather my plugin may break at some point. There doesn't seem to be a way around this for what I want to do. Is this ok?
No, as you say the plugin may break in the future and it won't work on windows IIUC. Ask on the dev ML for what you need in the API, it may be able to be added.
Is there another plugin out there that does this? I couldn't find one.
Would geany-plugins accept another plugin?
As Steven said, if you commit to maintaining it, most likely.
Cheers Lex
Thanks, Peter O'Malley _______________________________________________ Devel mailing list Devel@lists.geany.org https://lists.geany.org/cgi-bin/mailman/listinfo/devel
On 14-01-15 07:02 PM, Peter O'Malley wrote:
[snip]
- In the actual folding code I used sci_get_fold_level (and a couple
other scintilla wrappers) which aren't in the plugin API so I gather my plugin may break at some point. There doesn't seem to be a way around this for what I want to do. Is this ok?
You can just use the exposed Scintilla API, like:
scintilla_send_message(sci, SCI_GETFOLDLEVEL, line, 0);
Cheers, Matthew Brush