Remote server/TRAMP support (fix #84)

[FelipeLema]

Support remote files / tramp.

[joaotavora]

Hi @FelipeLema this looks good in general. But (as you probably already know)

• it is existing failing tests;

• it needs existing tests (how?);

• some of the function names are hard to understand;

• your formatting choices go against mine, so unless you feel very strongly about them, later on, I will ask you to review them.

[joaotavora]

it needs existing tests (how?);

Duh, I meant "new tests"

[FelipeLema]

it is existing failing tests;

yup, I'm using travis as checker for now. This is why I'm doing this as a /draft/ pull request. I wanted this dPR to be visible so no one would be doing the same work twice (visibility).

it needs existing tests (how?);

I'm aware of this. For now, I'm working on unbreaking tests. Once I have that, will work out a way to test this. The way I see it, re-running the same tests using /ssh:localhost;/some/path should be enough, so I might end up adding a meta-config.

some of the function names are hard to understand;

Please, point them out, so I can address them directly. Naming in this context is real hard.

your formatting choices go against mine, so unless you feel very strongly about them, later on, I will ask you to review them.

I noticed that some alignment went different, but it was unintended. I'll add a TODO for them.

[garyo]

Hope you don't mind someone chiming in on this. Your work might help my case: I run Emacs on Windows, but all the rest of my development environment is in Linux using WSL on the same machine. So it's not remote as such. Right now I'm installing the language server on the Windows side just so eglot and intellisense will work; I'd love to use the Linux-side language server, but there's a filename translation problem. Windows c:/foo/bar becomes /mnt/c/foo/bar in Linux (with default mappings anyway). If in the work you're doing here, you could allow for a filename-mapping hook (i.e. between Emacs's idea of filename and the server's idea), so I could use the Linux language server, people in my situation would be very grateful! (Note that it's no problem for Windows Emacs to start a Linux process using WSL, and get its stdio and so on. It's just the filename mapping at issue.)

[joaotavora]

Hope you don't mind someone chiming in on this

Not at all.

the work you're doing here, you could allow for a filename-mapping hook (i.e. between Emacs's idea of

This looks like a simpler addition, but at any rate a prerequisite for TRAMP support.

You should be able to achieve a bit of this hooking if you advice-add to eglot--path-to-uri and eglot--uri-to-path.

We could ask @FelipeLema to split his PR into two parts. One that supplies this indirection, and another one with TRAMP support that will surely use this indirection. However, if that confuses or complicates Felipe's TRAMP goal, I'd rather it waits a bit.

[FelipeLema]

I guess I could add some code for url and path treatment, but that might end up bloating up the code.

@garyo can you open an issue so we can discuss a (quicker) solution to your problem?

[joaotavora]

@garyo can you open an issue so we can discuss a (quicker) solution to your problem?

That's probably a good idea, yes.

[FelipeLema]

I have several reasons to not continue with this Pull Request

• I have eglot working with tramp, but inconsistently... it seems I'm missing some low-level config/way of doing things (program setup works OK with small tests, language server program is getting incomplete strings/bytes). It's definitely not easy to debug and find out just what.
• There are several tramp tools/functionalities from Emacs 27 not present in 26 that need to be re-written. I've tried this, but have been getting inconsistent results.
• When 27 arrives (at my system), I will pick up this feature. This branch/PR is mostly patching things to get it working in 26, so these changes are mostly noise with 27 available.