Per directory yaml - #1759

How would the sample .guake.yml file look like? And where could it be located?

@zeusintuivo right now guake.yml could be on any directory, when you land in the directory guake reads the file and extract the title.

I guess it could be extended or modified.

In my own machine, I have now some directories I frequently used with .guake.yml files with only a title, e.g.

title: "backend"

and when I get into that directory the tab has "backed" as title, it's very simple at the moment.

I see, this is an original idea. It could be even better if could perfom custom commands. Like set node version or rbenv or rvm or php version. Perhaps have a command line execution section. And also following that topic a autostart section. Something that would only run once vs command line execution that would run every time you go in the folder?

I am glad you like it ;) I am honestly just interested in the title. I am aware more things are possible as you said, but some seem a bit dangerous to me, or at least it seems it should the responsibility of a different component, like .bashrc or something like that.

Also, the API used for the proof-of-concept seems a bit poor to me honestly, and I would be afraid of adding too many options. Like is ok to set the name of the label twice in less than a second, but you could easily get into problems if you run a large command.

I would leave it just as a way to configure Guake on a per-directory basis.

Closing old PR

Hi, is there a way to reopen this and consider merging this feature?

You can reopen but please update the pull request description and the Reno chunk

Going back through issues, found this through an old issue and it looks like it could be nice, reopening and will check later. For now, you may need to rebase your branch to pass test cases.

Alright, I've gone and done some quick fixing up with the build errors so now we can look at the content of this. For one, we're going to need some docs written so people can actually know how to format the yaml files

can you confirm if this PR is still ok?

Pull request has good stuff but needs a little bit more written before it's a feature to add. We need docs so that people can know that this is a thing that guake will look for, maybe some nice controls in the UI for creating these files and then some testing.

I added a small explanation with an example in the Feature section of the documentation. I imagine we could use this file for more than just the title.

So far I have been using this for over a year and I love it, it makes it really easy to navigate my tabs, though I only get lost when I am in a remote machine via ssh. I haven't thought too much about that case anyway or similar: opening another shell from the original shell.

Great, I added the encoding. I am not sure how much effor to put on checking in the read_guake_yaml I would like this function to simply return a dataclass with the content of the file and maybe keep it cached somewhere in the parent class, until a change of cwd is detected. That class can be accesed then for some of its content, still some minor checks are required at every step where the content is intended to be used.

Also, I am using the extension .yml but I am unsure if everyone is used to have a `.yaml' instead.

I am happy to hear also if adding typings is in the future plans, fro vte in particular I guess I need to add an import on Vte and the vte argument is a 'Vte.Terminal' right?

Best.

The merge from master seems to be failing checks, I'll fix that real quick.

I would like to have description of this PR update at the OP description.

I still have to add a flag in the configuration to activate reading from this file. I will try to finish that during the week.

Regarding caching the content to prevent unnecessary reads, I honestly would first implement a very simple time() based control to prevent accessing the file if it has been accessed less than certain seconds ago. Would that be ok for you guys?

Yeah sure that'll probably be fine, generally should use monotonic() instead of time() though for the safety.

Back because build failed. For future reference, you can just run make black to handle formatting for you after you're done writing code, I'll make the check pass here for now. Also you can probably squash commits down if you have a couple with the same name

I added an options in prefs and a way to cache the content of the yaml files and prevent reloading too often. I honestly don't like to have the two properties _guake_yml and _guake_yml_load_monotonic but this is a simple solution I guess.

CI failing, you might need to rebase your branch onto latest git to resolve the weird dependency thing because I'm not seeing this issue crop up in any of the other active PRs. Looking at the things deepsource is complaining about, you can ignore the second one because it doesn't have anything to do with your changes but this one is relevant.

Hi @Davidy22 I did the rebase to master, but the error is still there. I honestly don't know what's the problem since make test runs for me. Could you suggest a way I can check on my end what's the problem?

The PR is now "ok" according to the last run. The tests are not passing in th CI, but they pass on my machine.

Regarding the code, I was planning to just add a FileManager class in utils.py to handle the cache for file content instead of the two properties in the Guake class. What do you thik about that?

Oh, I see you added a pyyaml line to the pipfile, so that probably changes the dependency graph and I suspect that's where the issue is coming from. You can do a pipenv lock to regenerate the lock file from your local setup if you have a working setup and that'll probably clear this up.