I'm looking at porting a pygtk application to Gtk 3 and gobject-introspection. When help is selected in the menu, the code calls gnome.help_display('appname') to display it.
The gnome package is firmly part of Gnome 2 - in Ubuntu it's part of python-gnome2, with lots of Gnome 2 dependencies. I can't find any equivalent package for Gnome 2. Is there any way to achieve the same functionality without depending on Gnome 2?
Apart from that function call, the app has no particular requirement for Gnome libraries. So a desktop-independent way of displaying the help, which is in Docbook format, would be ideal.
You can use Gtk.show_uri() For instance:
$ python
>>> from gi.repository import Gtk
>>> Gtk.show_uri(None, "help:evince", 0)
The first parameters is the Screen, the second the URI and the third a timestamp.
With respect to the documentation, I would recommend you to use Mallard, which is a way simpler than DocBook and it is oriented to build topic-oriented documentation.
Related
I'm writing my first KDE plasmoid using QML. The hello world example uses a metadata.desktop file, while this KDE Plasmoid tutorial talks about a metadata.json instead and says that the metadata.desktop is 'discouraged' now and a desktop file should be converted to json using desktoptojson.
However, when I browse the globally installed plasmoids under /usr/share/plasma/plasmoids/ they all have both the metadata.desktop and metadata.json.
First question: So, what is really recommended? Just the metadata.json? Or both?
And, I wasn't able to find the desktoptojson tool. I'm using Linux Mint and the ./kdesrc-build --initial-setup for debian based systems says that it's "This is woefully incomplete and not very useful" ... I read that "most users of this [i.e. desktoptojson] utility will use the CMake macro kservice_desktop_to_json as part of the process of building a plugin.". However, I haven't found the documentation yet how to use this.
Second question: In case one should maintain both files (for whatever reason), should I use desktoptojson to keep them in sync? And if yes, how?
Thanks!
First question: So, what is really recommended? Just the metadata.json? Or both?
In the current source code, most stock KDE applets such as the task manager use metadata.json's and have dropped the metadata.desktop's. It may be that the desktop files you have locally are left over from old versions, the new format was installed but the old one was never deleted.
Second question: In case one should maintain both files (for whatever reason), should I use desktoptojson to keep them in sync? And if yes, how?
The man page on Arch you linked to has all the information. The tool is part of the package kservice. Find the equivalent in the repository for your distribution. Then, to use it
as part of a CMake macro:
add_library(myplugin MODULE ${myplugin_SRCS})
kservice_desktop_to_json(myplugin myplugin.desktop)
directly on the command-line:
desktoptojson -i myplugin.desktop -o myplugin.json
I have to make some functions that will use different lanaguages (python, R, js).
I got stuck at the part of generating random numbers in Python to initialize a list with random elements. I looked up on ways of initializing random lists, and then I decided to use result = polyglot.eval("python", "[random.randint(0,10) for i in range(20)];").
The problem that I face now is that I need to import the "random" library from python, or whatever libraries will I need from different languages. I heard that it might be a problem with the dependencies, but I am not sure...
What am I supposed to do? Is it even possible to import libraries from more languages in a single project? What other alternatives do I have?
Note that solution for different dynamic languages may differ.
Also js component is stable, while python (as of 2021) is still experimental.
Here is example for Python with modules
https://github.com/paulvi/java-python-graalvm-template
And if you really do polyglot (using Python object in Java code),
see https://github.com/hpi-swa-lab/graalpython-java-example
There is still issue how to actually deploy this in production
https://github.com/hpi-swa-lab/graalpython-java-example/issues/6
as just bundling venv subfolder into jar, will just work.
One solution is in ttps://github.com/paulvi/java-python-graalvm-template
Also randon, i.e. any library with graalvm is still big issue, as different packages have different issues, see https://github.com/oracle/graalpython/issues/228
I suggest, that before really mixing a lot of languages, just try one, e.g. js that is more stable, make it work, and then try next.
BTW PyCharm does not yet support graalpython.
If you do any open source, or later find somethin new, please let me know via GitHub issue
I have already built pypi package stored on pypi server few days back. Now I want to compare source code diff between already built pypi package and recent code built today. Is there any way to this?
I want to compare already built pypi package and newly build code. And If there is any difference in source code then only create a new package and upload it to pypi server
If you have only Python bytecodes, you cannot get the corresponding source code (that hypothetical transformation is called decompilation, and is not possible in general; read e.g. about Rice's theorem). Since any translation (such as the one done by the python program) from source code to bytecode is losing some information (e.g. name of local variables, comments explaining the intent of the code).
Equality of the behavior of functions by static analysis of their source code (and the observable behavior of your code is what you really care about) is an undecidable problem. Learn more about λ-calculus, it is deeply related to that question.
The source code (by definition, the preferred form of code on which developers work) is not only for computers, but mostly for fellow developers: in other words, most of its value and its meaning is a social one (and that is what free software is about). Read more about the semantics of programs.
For example, renaming a variable from i to x may convey the implicit hypothesis that the intended dynamic runtime type of the value of that variable was an integer, and becomes a floating point.
Maybe you want some kind of package manager (or some version control system, if you deal with source code, or some build automation tool, if you build then install software out of it). Python has something to manage packages. The scons build automation uses Python, but there are many other build automation tools, GNU make being a common one (that you could use to drive compilation from .py source files to .pyc bytecode files and their installation). For version control, I recommend git.
PS. Your question is very unclear and smells like some XY problem.
I am using pandas to do some data analysis. Others in my company are wanting to process data in a similar fashion, but won't want to use a programming language to do it. After significant googling, I found Orange, which has the perfect interface for what I'm wanting people to do. However, the widgets don't do the types of tasks we're looking at. So, I decided to see if I could write my own widgets for Orange to do the tasks.
I'm trying to use Orange3; this seems like the best bet when I'm using WinPython. I must say that going through the documentation for widget creation (for Orange2) and the code for the Orange3 widgets is rather impressive - very nicely written and easy to use to implement what I'm wanting to do.
After writing a couple of widgets, how do I get them into Orange3? the widget creation tutorial is for Orange2 (in Python 2.7), and I haven't got it to work for Orange3.
My project is at the moment rather small:
dir/
orangepandas/
__init__.py
owPandasFile.py
pandasQtTable.py
setup.py
setup.py currently contains the following:
from setuptools import setup
setup(name='orangepandas',
version='0.1',
packages=['orangepandas'],
entry_points={'Orange.widgets': 'orangepandas = orangepandas'}
)
When I run python setup.py install on this and then try opening Orange3 canvas, I don't see my shiny new widget in its new group.
After tracing through how Orange3 imports external libraries, it seems that Orange relies on the actual widget file existing, rather than being inside a egg (zipped) file. Adding
zip_safe=False
to the setup options allowed Orange3 to import the widgets correctly. Orange3 uses os.path.exists in cache_can_ignore in canvas/registry/discovery.py to detect if the path exists at all, and if it doesn't, it doesn't try to import it. Using zip_safe=False makes sure that the addon stays uncompressed so that the individual files are accessible.
(For the next person who tries to do what I was doing.)
Is there a way to export all symbols in a Julia module (something like the semantic counterpart to importall)? Such a functionality would be very useful when the number of symbols to be exported grows large. TIA.
There's the Reexport.jl package that provides a form of this; it's use case is when you have a submodule and you want to reexport all of the exported symbols from the inner module into your current module. I know, not exportall, but part of the functionality.
I wrote a minimal ExportAll.jl package. It works and is tested for Julia 1.1.1 and it is available as a Julia package. Basically, it exports all symbols(!) defined in a module that is not part of any of the core modules. Disclaimer doing things like this is not good practice and should only be done with great care.