Is there any "pre defined libraries" present in progress4GL?Like Math.h in C.
There are a number of useful libraries at oehive.org.
There are also some interesting things in $DLC/src. ".pl" files are "Progress Libraries", they contain collections of files that you can run so long as the .pl is in the PROPATH.
Comprehensive documentation of what can be found where is "weak".
You can link in an outside library, but outside of that - no. Support for all the language elements are already built-into the virtual machine that runs the ABL code.
Related
I'm looking to improve my own productivity by trying to figure out how to provide cmd+click functionality for variables declared at runtime. I work with a library that allows you to register objects to it at runtime to be accessed later — it would be great if I could cmd+click to definitions of these! How do you think I could achieve this? I'm open to any ideas. Thanks!
Edit:
Thoughts I have so far inspired by feedback from various Slack communities:
There is no such thing as 'runtime' when providing functionality in IDEs
We need to be able to teach the IDE about what definitions to expect. Whether this means writing an extension to use the VS Code API specifically for my use-case, or otherwise working within a system that generates files that the IDE already knows to look for, which allows it to connect the dots and provide the precious peek / goto definition functionality
Edit:
I'm starting work on a VS Code extension that reads a dotfile for configuration — I'll share what I find out as I go!
I did my best to overcome this problem but it was in vain. I am developing tests in Robot Framework (RF) in Visual Studio Code and I want it to either show "signatures" of keywords from standard libraries or to be able to go to their definitions. Go to definition and signature displaying when hovering with Ctrl key are behaving this way:
works for my keywords written in RF (hoorah!)
works for my keywords written in Python only if a keyword consists of one word (I think conversion between undescores and spaces is failing)
does not work for keywords from standard libraries even if the keyword is one-word (e.g. "Fail"), regardless whether the keyword comes from built-in libraries or other ones (e.g. SeleniumLibrary)
When failing, Robot Framework Intellisense Server gives me message "Keyword definition 'Blah Blah' not found from the workspace".
I am using plugins Robot Framework Intellisense FORK and/or Robot Framework Language Server. I tried to configure them carefully according their documentation (Details), but the best state I reached is described above.
Can you help me please? I do not need to use a specific plugin, I just need to have the signature or documentation (or implementation) of every keyword in my code to be one click far.
According to their documentation they support
Goto definition
For variables
For user keywords
I was able to partially solve this when I not only open the folder with source code, but create a workspace and add the folder containing standard libraries to it as well (Add Folder to Workspace). I do not know exactly which folder is the best to add, it seems ...\Python\Python37\Lib\site-packages\ to me.
I have RestClientLibrary and UserFunctionsLibrary
UserFunctionsLibrary needs RestClientLibrary in order to function.
When I compile these down to libRestClientLibrary.a and libUserFunctionsLibrary.a how will they be able to interact with each other?
In Xcode currently I have set the User header search paths to find the .h files and I have linked the UserFunctionsLibrary with the RestClientLibrary binary. However, when distributed other users of these libraries may have different set ups and such. I can't see that it will work.
Thanks for any insight you can give me.
Those .a files are just library files. They will need to be linked together to actually be used. The linker will handle resolving all the symbols from RestClientLibrary into UserFunctionsLibrary.
As far as other users, they will have to configure their system in a way such that both libraries are passed to the linker.
GTK can construct images by name of the "icon from current icon theme". For example:
#!/usr/bin/env python
import gtk; wnd=gtk.Window(); img=gtk.Image();
img.set_from_icon_name( "go-jump", gtk.ICON_SIZE_BUTTON );
wnd.add( img ); img.show(); wnd.show(); gtk.main()
This will display a window with a pretty arrow inside it. But - only on ubuntu. On Windows or OSX it will display window with a "no image" icon inside :(. So my questions is - where GTK keeps icon names like "go-jump"? Is it some list or specification available like for stock icons? Maybe it's some GTK API i can use to list such icons?
The names are in the Icon Naming Specification. If this doesn't work on Windows or OSX, then report it as a bug - GTK needs at least one icon theme installed in order to work properly.
Better late than never !
Here are two little Python scripts showing all existing icons sorted by their name:
http://dl.0xdeadc0de.fr/misc/gtkiconview_freedesktop.py for those in the Icon Naming Specification.
http://dl.0xdeadc0de.fr/misc/gtkiconview_local.py for those included in packages such as 'gnome-icon-theme-symbolic'.
Developing on Debian but wanting cross-platform support, I've recently installed gtkmm and co on both Windows and OS X, via MSYS2 and Homebrew respectively. As an aside, gtkmm, MSYS2, and Homebrew are excellent projects and well worth a look (I have no affiliation blah blah).
In that time, I think I've gained a decent understanding of why this happens and how to 'fix' it.
Why
This is not a bug in GTK+, as the accepted answer suggests. It's a fact of life when using a library tailored to Linux on another platform, via an environment that creates a Linux-like picture.
On Linux, you have a single environment with a standardised filesystem - via the FHS and FreeDesktop specs - with icons always in /usr/share/icons.
In contrast, on other platforms, you install Linux-like environments for compilation and sometimes runtime, each with its own virtual filesystem - which is difficult to track by other programs, without doing dangerous things to the end-user's PATH, especially if you're trying multiple such environments... You get the idea. The end result is that your program, by default, doesn't necessary have any awareness of where your environment's virtual root resides.
How
The solution, I've found, is to copy all required resources to the same folder as your executable - i.e. the directory you'll eventually redistribute, or if you're writing OSS, do this in your makefile or such. This is because, as far as I can tell, a program - looking for GTK+ resources, DLLs, etc. - checks its own directory before the (probably not helpful) PATH. A little opposite to how shells do things, but handy!
So, you copy the relevant icon themes, as installed by your package manager, into a new folder within said directory. At a minimum, do this for the standard theme: copy it to $YOURDIR/share/icons/Adwaita. Then restart your program. If you're anything like me, now all the icons work perfectly.
The same applies for pretty much any other resource that you need to redistribute with your application.
Notably, I'd suggest doing the same thing with all DLLs needed by your binary. These may or may not work for you out-of-the-box and depending on how you invoke your binary - but you really can't afford to speculate about whether your end-users already have the right set of DLLs installed somewhere in their PATH, of the right version compiled by the right compiler.
Can Emacs support the concept of projects? That would be where several configuration options varies depending on what project you are working on.
Yes, please look to the EDE package from CEDET library
Check out the project-local-variables plugin. I haven't used it personally, but I've seen it mentioned in the Rinari-mode documentation for scoping down a find-file-in-project function.
You generally wouldn't expect emacs to know about differences in build requirements---you let the build system (make or whatever) take care of that. To handle different coding standard define local yourproject.el files to set variables, load and unload local modes, etc... (use M-x load-file to run it).
To be pedantic, Emacs can support anything, being as it's a Lisp language runtime which so happens to have an editor on top of it.
To be specific, some addons for Emacs support Projects, such as the JDEE.
Well, as you can see there are several packages to do this.
You might like this approach:
http://blog.jrock.us/articles/eproject.pod
I just read it today.
Dude, as mentioned in another answer, EMACS can do anything. That said, let's expand a little on what a "project" means in this context. In most IDE's, like Eclipse, Netbeans, Visual Studio, or Xcode, "project" means two things: a file layout for the pieces that belong together, and instructions to the build system to build the thing composed of those pieces.
In EMACS, you have complete control of these things, but how they work together is dependent on both the target environment -- Lisp, Java, C/C++ and so on. Lisp is supported by SLIME, Java by the JDEE, C/C++ by the built in modes and a bunch of conventions; version control is also built in if you want to use it.
In most cases, the build details don't really affect the editor itself much, but you can specialize the behavior for specific files using either modelines, or local variables. JDEE has a specific "project" notion built in.
What languages and environments are you trying to work with?