I know this comment is a lot of words, and may feel critical, but I assure you I mean them in the spirit of "for your consideration" and want to start off by saying thank you for sharing such a cool toy with the world under a permissive license. Great job, and thank you!
I was trying to find out which version of python it offered, and while digging into that I found surprising references to win32 binaries in the pyproject <https://github.com/Amourspirit/python_libre_pythonista_ext/b...> although the extensions page says it's for all 3 major OSes.
Related to that permalink, please do consider creating a formal tag that represents the code that went into the 0.1.1 binary you linked to. It'll greatly help those trying to track down bugs if they don't have to $(git bisect) in order to find out which sha created the extension they're using
Kind of related to that, future you (and folks who clone your repo) are going to be sad if you keep putting the release artifacts in git, since it will make your repo grow without bound. If you want to make the .oxt available to folks outside of the libreoffice.org URL, that's another fine reason to create a tag since GitHub will cheerfully hold on to the .oxt with a permalink forever, but outside of your repo. It'll also motivate you, or a contributor, to create a GitHub Action showing how normal people could possibly build the release artifact for themselves
shouldn't this functionality be included in the base libreoffice itself? so... just asking, how much efforts would it take to integrate it fully into the code as opposed to an extension? that way more people would use this feature and more people would help improving the code.
good job. need more work done like yours in libreoffice.
Yes, some LibreOffice developers commenting on this agree with you. We also have existing proposals like "Ship Python package manager (PIP) with LibreOffice internal Python" https://bugs.documentfoundation.org/show_bug.cgi?id=162786
I have been meaning to use Grist for an educational org I volunteer for - mainly to manage student data. We would definitely self-host, and it is great that there is already a Class Enrollment template. Your self-hosting docs [1] seem pretty straightforward. However,
* I am annoyed that the the app requires providing inputs through command line args or environment variables rather than a fixed config file that I can store in git.
* Some directions on creating a service that starts the app on server restart etc would be good.
Thank you. I've been struggling with a good story for configuration. I'm not entirely satisfied either with using environment variables for everything.
Using Docker, you can just put your env-vars in a file and execute Docker with --env-file specifying where your configuration is, which is what I currently do, but perhaps a .env file would be more standard and comfortable.
Let me put it this way. Your app has wide possible usage, including a lot of casual usages, by relatively non-technical users. Like small businesses or clubs etc, as your templates show. Spreadsheets run the world, and grist is turbo spreadsheets.
If you want those people to use your software, you should make it as easy as possible to install it. Think of a tech-minded teenager helping their dad set up grist to track inventory for their corner store. How can you help them install the software in a couple of hours?
It is intended to replace spreadsheet use, and the database columnar restrictions are intentional.
Moreover, since it's using sqlite in the backend for the database, cells are allowed to hold the wrong type (and big, red warning colours light up when this happens). This is a sqlite characteristic. So while it mostly does enforce database-like behaviour, it also allows spreadsheet-like lax behaviour.
It really is a mixture of database and spreadsheet.
Not until someone implements the handling of such in XLSX import. Excel needs a cloud connection to run Python and there are apparently some homegrown functions like xl(), so it probably would not be straightforward:
Why in gods green earth would you want a cloud connection to run a Python script on your computer when you canr run python locally since day 1 of Python ?
It's AI. Have no background information you don't have I think I'm just good at spotting it.
Scales are intricate and seem correct yet the gears are irregular and often not gear shaped, assymetrical etc.
Some things look nonsensical and functionless. The circle around has has imperfections and the resolution of the image is too low.
Artists are all-too often terrible about drawing gears/mechanical things.
EDIT: That said, I agree it's likely AI --- the odd mis-matches of scale size, and the lack of interest in how the background elements interact are what I would view as signs of AI generation.
That said, I'd be curious about which tool was used to make the file, and what the prompt was --- I really wish that there was a way to embed that persistently in the underlying file so that it could then be revealed --- or, a regulation that all AI-generated images be uploaded to a central repository where, when image search reveals them, the specifics of their creation are made known.
> A real artist already putting this much effort into it, would never make this mistake
I'm not so sure that's true. See https://youtu.be/6JwEYamjXpA?t=321 Granted - most of the examples in Matt Parker's talk isn't exactly work of art, but I'm sure effort were put into them.
Edit: I agree the way the cogs don't work in the project's logo is of a different class, so maybe you have point.
Definitely AI. I have a weird visceral reaction to AI generated images and that triggered it.
(I put it down to distracting myself with an AI image generator while I was very, very unwell and now my brain equates the output of one with t'other).
I feel like if you're asking how to shove Python into a spreadsheet app, you've made a mistake somewhere in your process and should reconsider everything.
Walk us through your hypothesis. People solve problems in various ways, and workflows can change. Having more options can be useful and lead to new solutions.
Microsoft offers Python for Excel for quite some time now [1]. Seems like a natural fit, when you are used to working in spreadsheets and want to add the low code machine learning capabilities that Python offers.
I'm in the process of ripping gsheets out of some internal processes, and the pain they are causing me are immense, so I understand the sentiment of this.
However. If those sheets hadn't existed earlier on, I might not have a job at all because they business wouldn't have been able to easily/cheaply solve the problems that they did
