Consider how to ignore a sheet

[jamesaoverton]

Currently, if you cogs add two tables and then cogs push, the Google Sheet will end up with just two sheets, and all other sheets will be deleted. (Note that you can revert changes using Google Sheets history.)

https://github.com/ontodev/cogs/blob/master/cogs/push.py#L215

Now I have a Google Sheet with a bunch of formulas to summarize data on other sheets. COGS doesn't handle formulas, so I can't cogs add this sheet. I don't need a local copy of this sheet at all. But I don't want COGS to delete this sheet when I cogs push. I'd be happy if COGS just ignored it.

It would be simple enough to provide a list to clear_remote_sheets that it would ignore. Then I guess we could use the same list to ignore sheets for cogs fetch and cogs merge (#102). Maybe we need a new cogs ignore/unignore command, or just an --ignore option for cogs add/rm? And I'm worried that I'm missing other problems, or just making the mental model too complicated.

[beckyjackson]

I think cogs ignore makes the most sense. Maybe instead of cogs unignore we can just do cogs ignore --revert or something.

I guess this could go in a config file - something like .cogs/ignore.tsv?

[jamesaoverton]

Let's try adding an ignore column to .cogs/sheet.tsv and use cogs ignore --revert.

[jamesaoverton]

We also need to think about other commands:

• cogs ls list ignored sheets?
• cogs status no change?

[jamesaoverton]

Now that I've tried #110 and had a bit more time to think about it, I'm still not completely clear on the right behaviour.

1. Now I think that we shouldn't (ever?) delete sheets from the Google Sheet. So I guess sheets would be ignored by default, until cogs added.
2. If sheets are ignored by default, then you would only use cogs ignore to stop tracking a sheet that you previously cogs added (without removing it). Then maybe we don't need cogs ignore but just cogs rm --ignore option?
3. Now I think that the opposite of cogs ignore is cogs add.

[beckyjackson]

In this case, do we even need an "Ignore" column? If the default behavior is to just ignore any remote sheets that are not in .cogs/sheet.tsv, that seems redundant. Can you think of a case where it may be useful?

[jamesaoverton]

I was thinking that .cogs/sheet.tsv would still have a row for each sheet in the Google Sheet, and then that would still be useful for cogs ls or whatever.

[beckyjackson]

Gotcha, that makes sense.

I think we can get rid of cogs ignore. I don't think I like the cogs rm --ignore option because that feels like you're deleting the remote sheet. I would rather do something like:

cogs add sheet_title

Where usually we require a path (cogs add PATH), if the provided title (instead of path) appears as an ignored sheet in .cogs/sheet.tsv, we will just stop ignoring it.

[beckyjackson]

I updated the PR. New remote sheets are now automatically ignored. You can "unignore" a sheet using add. From the docs:

Adding an Ignored Sheet

By default, COGS ignores any untracked remote sheets. If you wish to start tracking a new remote sheet, you can do this by passing the sheet title to add:

cogs add [title]

The default path for pulling changes from this newly-added sheet will be the current working directory (the same directory as .cogs/ is in) and will be a lowercase, space-replaced version of the title (e.g. My Sheet becomes my_sheet.tsv). If you already have a tracked sheet at this location, the date & time will be appended to the path (e.g., my_sheet_20200922_103045.tsv for a sheet fetched at 10:30:45 on 2020/09/22). This path can be updated with cogs mv.