search

3rd / image.nvim

🖼️ Bringing images to Neovim.
MIT License
812 stars 35 forks source link

update images on modifications #137

Closed jonboh closed 3 months ago

jonboh commented 3 months ago

Currently image.nvim does not have a way to detect that an image changed since the last time it was rendered. This PR should address that case. I've added added a check for the date:modify property which will clear the image and force a re-render when it has been modified since the last render. It seems that the initialization of self.resize_hash was duplicated so I've renamed one of them to self.date_hash. I've modified the cache and tmp files generated, which will now also have the modification date of the file.

Here's the behavior of the master branch right now: https://github.com/3rd/image.nvim/assets/31407988/c258801f-ac85-491b-a7b3-e2dadcefb0ad Note that it is necessary to open and close nvim in order for it to render the updated image (I think closing the buffer and reopening also works)

And here the behavior with these changes: https://github.com/3rd/image.nvim/assets/31407988/c43c1e1f-b658-4d0b-8f8e-b804adfb3b55

3rd commented 2 months ago

@jonboh reverted this because it was killing performance, probably because of the constant magick load, could be worked around by checking the file modification time, exif isn't really relevant anyway

jonboh commented 2 months ago

Ok, do you have a workflow/scenario in which it was specially noticeable? I would like to try and find a fix for both the updates on change and the performance. A different approach could be to provide the user with a way to refresh all (or a specific) image.

jonboh commented 2 months ago

I've tried using stat -c '%Y' with io.popen to check the date, and with an image on each line there's a noticeable performance drop (even worse than with magick get_property I think).

Maybe it would be ok to add a flag for this update behavior (similar to only_render_image_at_cursor), and to limit it to just upadet the image under the cursor so check_image_changes_under_cursor (otherwise don't check anything, so the current behavior). In this case I don't think the performance hit from loading a single image would be noticeable. What do you think? I can prepare a PR with these changes if you think that would we acceptable

jonboh commented 2 months ago

I've implemented the change in https://github.com/3rd/image.nvim/pull/150 Let me know if you have any issue with the solution @3rd :)