[concept] plugin launcher

driven

@Jim,
did you ever get the default text?
what happens if you type into the box?

john

Jim

@Jim,
did you ever get the default text?
what happens if you type into the box?

john

LaunchUp worked when I first tried it a few weeks ago - now a get the dialog which only shows me the source for the html. No idea what I did.

driven

@jim said:

now a get the dialog which only shows me the source for the html. No idea what I did.

I would check other WD's first to see if it's a 'LaunchUp' specific problem

If you haven't touched the files, I'd look at the IE settings.

have you disabled js or set it to 'show as plain text'.

can you set those type of things a PC?

sorry, not very helpful

john

Aerilius

I have never seen this problem.
What happens when you create a webdialog on the console and then set_file(".../LaunchUp.htm") ?

I don't believe that, but could there be a reason that Internet Explorer thinks it's a plain text file and not html? (confused mime type, or is the encoding or file header wrong?)

Jim

When I inspect the html in the dialog, I see it is incorrectly nested somehow..

<HTML><HEAD></HEAD>
<BODY><PRE>&lt;!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/strict.dtd"&gt;
&lt;!--
This file is required by LaunchUp.rb"

Copyright 2011, Andreas Eisenbarth
All Rights Reserved

Aerilius

I'm not sure what is nested wrong? (I haven't found it yet)
I know many browsers internally prepend a <HTML><HEAD></HEAD><BODY><PRE> when they display plain text files, so maybe this is only the result instead of the cause why IE shows the plain text.

I found that the doctype makes no sense (transitional <-> strict). Can you try
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">

Jim

I tried the new doctype - nope. I got rid of the doctype and comment beginning the .htm file with <html> and it started working again.

I put the doctype back in and removed just the comment - it worked.

driven

I've got v2 stable enough to start looking at it on Safari... Selection from the search results only works with new keycodes and not the mouse...

HTML Changes include:
added a break after 'form' and moved the js to below that break... so line numbers reflect this.

Line 01- DOCTYPE html
Line 22- added to body>> height:100%;
Line 23- changed in form>> top:4px;height:inherit;
Line 49- input type="search"
Line 57- [<]br /[>] ...without []
Line 59>>296- scripts
Line 210/211- // left-arrow: select first entry
if (keycode == 37) {
Line 222/223- // right-arrow: execute entry
if (keycode == 39) {
Line 275- // $("combo-input").focus(); #jb_mac

in the Rudy:
this forces it to only scroll down
Line 54- changed to @launchdlg = UI::WebDialog.new("LaunchUp", false, false, 0, 1140, 270, 50, true)
Line 92- added @launchdlg.set_position(1140,0)
to get it to show_modal
Line 98>>102 + Line 110>>102- if RUBY_PLATFORM.include?('darwin') # Do NOT 'do this if on a PC! @launchdlg.show_modal else @launchdlg.show end #mac _modal

I also set a shortcut key in 'system' preferences and have started updating the txt file.

john

Aerilius

Thank you so much for your efforts! I included it in the code. For me it used to work in my browsers/configurations, so otherwise I couldn't have found these solutions on my own.

I did some search about comments between doctype and <html>, but didn't find anything about issues. I probably leave the comment completely away (or does it work with the comment after <html> ?)

I was a bit hesitant about %(#000000)[input type="search"] because it's not standard (html4), but since unknown type defaults always to text, it shouldn't be a problem and doesn't break anything in Internet Explorer.

Dan Rathbun

@jim said:

I put the doctype back in and removed just the comment - it worked.

It IS a bit weird that the comment is outside the <html> block... I would put it inside the <head> block as well.

driven

Hi all,

I'm bumping this, because I want to modify it to work with 'on_off' http://forums.sketchucation.com/viewtopic.php?f=180&t=41994#p372262

Using the option to open SU without rubies but with Ruby we have the ability to open individual .rbfiles or folders containing .rb files using Alex's + Jim's methods or load from RC, which is great.

However if you want TIG's 2DTools [for example] you need to change all the 'requires' in all the scripts into 'load' + full the path before it will run with 'rubies' turned 'off'. I've got that working and it will still run as normal with them 'on', but would be a pain to do for each new version for all my rubies. I haven't worked out any of Fredo's yet, but I'm sure there's a way.

So I'm thinking it would be a lot easier to parse the plugins each time you call one to 'load' and leave the originals intact,
if it needs support files they can be added, but also need full path's.

Basically, use 'LaunchUp' to select the 'standard' ruby rather then it's individual items,
because you only need to have a few sets on at a time the toolbar count is still low.
when you get to many you start again.

I'm stumbling through so any feed back is appreciated.
john

Aerilius

Hi,
This is very far from a proper plugin. Parsing other ruby files in order to get commands is not an ideal solution and needs always manual editing. I would not publish the current state as a plugin.

I've played around with alternatives (should probably make my current version ready and post it here). I tried aliasing the UI::Command before all plugins load. Problem is that this way I modify core methods, I have to make sure that my plugin loads always first and it can be harmful if another plugin aliases UI::Command.

Jim inspired me to get the commands with Objectspace, which works for all the command's metadata, but I haven't figured out how to call the proc.

Third possibility is to continue hoping and begging for SU9 that we will have better access to commands (and maybe removing/modifying toolbars and menus).

lapx

This idea deserves an award! I spend too much time trying to find certain commands and dealing with grey out issue.
Is it necessary to create another plugin folder or can I just check not load at start up.
Have'nt tried yet but appears to be a winning concept if it can work smoothly.

driven

Hi Aerilius,

thanks for replying

I may have found a workaround for what I'm after using Fred's '000_AdditionalPluginFolders.rb' with 'on off.rb',

I've made and registered some subfolders in SU, then when I launch with 'rubies OFF' and run '000_AdditionalPluginFolders.rb' I get asked which I want to instal, so I can choose 'Base Set', '2D', 'Fredo' or 'TIG', etc...

This avoids having to look at or modify the rubies content at all and with a simple folder creator/sorter WebDialog this is almost doing what I want [on mac at least] and without any .rbs [but they can be accessed via a full startup].

I'll keep digging
john

Jim

@aerilius said:

but I haven't figured out how to call the proc.

Once you have a reference to a proc, call call on it.

pr = Proc.new { puts "Hi" } pr.call Hi

Aerilius

But can I get the proc from Objectspace? I didn't get it to work...

Jim

@aerilius said:

But can I get the proc from Objectspace? I didn't get it to work...

Oh yeah. You can get UI::Command objects from ObjectSpace, but you can't get a handle to a Command's associated Proc.

You can get Proc objects from ObjectSpace

ObjectSpace.each_object(Proc) {|pr| puts pr }

But there is no handy way to know which command a Proc is associated with other than the String representation (which is the filename and line number where the command is created.)

So no "good" solution. I have asked Google for a method to access the Proc of a Command. That would make things sooo much simpler.

dukejazz

Your idea of a launcher with a quick search, is something Im looking for. Type a search for any plugin/tool that you want to execute. If the plugin is not yet loaded, it will be loaded instantly.

"The basic purpose is to keep only a small range of standard plugins (=> quicker start of SketchUp and uncluttered interface), but make it easy to load rarely used plugins when you need them. You don't have to bother anymore if a plugin is disabled/archived or not."

You created a prototype. It uses an index file that contains a list of info.
This is your primiss. I did tryed your plugin and its a qood ideal.

There are several things your are worrying about:
" 1) It should be at least techinically possible to load any existing plugin that is disabled"

2)"But how to get the commands that should be executed, a proper name, icons, eventually keywords?" Well this is to much work for me?. I try to useing SUTool: a new way to manager your RB script but it dosn't load every thing. You have to double buffer everything "(since some plugins contain complex structures of classes and wrapper methods, not to speak of .rbs plugins)"
SUTool:
These were my ini files
01file.txt
02edit.txt
03toolbars.txt
04view.txt
05camera.txt
06draw.txt
07tool.txt
08plugins.txt
09context.txt

(Stage 1)
Is there a better solution? yes window/preferences_shortcuts,a dummy key, with a macro.ank, it would give API access to all commands created with UI::Command,

(Stage 2)

I will tell you how at My tips about running more than one sketchup
http://forums.sketchucation.com/viewtopic.php?f=79&t=40254

thomthom

@jim said:

@aerilius said:

But can I get the proc from Objectspace? I didn't get it to work...

Oh yeah. You can get UI::Command objects from ObjectSpace, but you can't get a handle to a Command's associated Proc.

You can get Proc objects from ObjectSpace

ObjectSpace.each_object(Proc) {|pr| puts pr }

But there is no handy way to know which command a Proc is associated with other than the String representation (which is the filename and line number where the command is created.)

So no "good" solution. I have asked Google for a method to access the Proc of a Command. That would make things sooo much simpler.

I have a proof of concept that intercepts the creation of Command objects and make everything available. Even the procs. I hooked into menu and toolbars as well so I could traverse them in a tree structure. I was able to do a string search of the commands, much like the described concept of OP. But I haven't released it as it would need testing and a robust error handling.

Aerilius

I had experimented with intercepting commands, but I found it too risky for me because I don't have enough Ruby experience. When there are two scripts intercepting the same methods, it would crash SketchUp.

Nice to see that this idea is becoming more common, it isn't really special. If only SketchUp made it's commands easierly accessible...

http://www.markshuttleworth.com/archives/939