html_url,issue_url,id,node_id,user,user_label,created_at,updated_at,author_association,body,reactions,issue,issue_label,performed_via_github_app
https://github.com/dogsheep/twitter-to-sqlite/issues/39#issuecomment-607010634,https://api.github.com/repos/dogsheep/twitter-to-sqlite/issues/39,607010634,MDEyOklzc3VlQ29tbWVudDYwNzAxMDYzNA==,9599,simonw,2020-04-01T03:45:16Z,2020-04-01T03:45:16Z,MEMBER,"OK, fix is applied to everything now.","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",590666760,--since feature can be confused by retweets,
https://github.com/dogsheep/twitter-to-sqlite/issues/39#issuecomment-607003655,https://api.github.com/repos/dogsheep/twitter-to-sqlite/issues/39,607003655,MDEyOklzc3VlQ29tbWVudDYwNzAwMzY1NQ==,9599,simonw,2020-04-01T03:18:00Z,2020-04-01T03:18:00Z,MEMBER,I've got this working for the `user-timeline` command.,"{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",590666760,--since feature can be confused by retweets,
https://github.com/dogsheep/twitter-to-sqlite/issues/39#issuecomment-606850453,https://api.github.com/repos/dogsheep/twitter-to-sqlite/issues/39,606850453,MDEyOklzc3VlQ29tbWVudDYwNjg1MDQ1Mw==,9599,simonw,2020-03-31T20:14:58Z,2020-04-01T03:03:50Z,MEMBER,Actually I'll hard-code the population of `since_id_types` to get known ID constants.,"{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",590666760,--since feature can be confused by retweets,
https://github.com/dogsheep/twitter-to-sqlite/issues/39#issuecomment-606998669,https://api.github.com/repos/dogsheep/twitter-to-sqlite/issues/39,606998669,MDEyOklzc3VlQ29tbWVudDYwNjk5ODY2OQ==,9599,simonw,2020-04-01T02:57:36Z,2020-04-01T02:57:36Z,MEMBER,"The tricky thing here is thinking about the interaction between the recorded since_id and a desire to run the initial import.

The first time you run `twitter-to-sqlite user-timeline db.db username` we want to fetch as many tweets from that user as possible - probably around 3,200 before the API limitations cut us off.

We need to record the maximum ID from those as the `since_id` - which we will see on the very first page we paginate through. That way next time we run the command with `--since` we will only fetch new tweets.

But what happens if our initial import is cancelled after only a few tweets? We risk never pulling in the rest of the tweets.

Not sure if I need to solve this at all or if I should instead trust users to run the command a second time without `--since` if they think they didn't retrieve anything the first time.

I had considered letting `--stop_after=` over-ride `--since` but that doesn't actually make sense - if you send a since_id to the Twitter API you'll never get back more tweets than exist after that ID, so the `--stop_after` would not make a meaningful difference.","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",590666760,--since feature can be confused by retweets,
https://github.com/dogsheep/twitter-to-sqlite/issues/39#issuecomment-606850008,https://api.github.com/repos/dogsheep/twitter-to-sqlite/issues/39,606850008,MDEyOklzc3VlQ29tbWVudDYwNjg1MDAwOA==,9599,simonw,2020-03-31T20:13:59Z,2020-04-01T00:23:00Z,MEMBER,"Table design for `since_ids` table:

type | key | since_id
--- | --- | ---
1 | 124324 | 2347239847293
2 | 99ff9cefff5cbfd804f7cd43e2b27ced8addbe8d | 2125947927344

Primary compound key on `(category, key)`

`type` is also a foreign key to a `since_id_types` table with `id` and `name` columns (probably created using https://sqlite-utils.readthedocs.io/en/stable/python-api.html#working-with-lookup-tables )","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",590666760,--since feature can be confused by retweets,
https://github.com/dogsheep/twitter-to-sqlite/issues/39#issuecomment-606843224,https://api.github.com/repos/dogsheep/twitter-to-sqlite/issues/39,606843224,MDEyOklzc3VlQ29tbWVudDYwNjg0MzIyNA==,9599,simonw,2020-03-31T19:59:11Z,2020-03-31T20:06:32Z,MEMBER,"Or... have a single `since_ids` table to track since values, and have its primary key be a string that looks something like this:

`user:123145`
`home:23441`
`mentions:23425`
`search:99ff9cefff5cbfd804f7cd43e2b27ced8addbe8d`

That last example would use the hash generated here:

https://github.com/dogsheep/twitter-to-sqlite/blob/810cb2af5a175837204389fd7f4b5721f8b325ab/twitter_to_sqlite/cli.py#L792-L808","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",590666760,--since feature can be confused by retweets,
https://github.com/dogsheep/twitter-to-sqlite/issues/39#issuecomment-606844521,https://api.github.com/repos/dogsheep/twitter-to-sqlite/issues/39,606844521,MDEyOklzc3VlQ29tbWVudDYwNjg0NDUyMQ==,9599,simonw,2020-03-31T20:01:39Z,2020-03-31T20:01:39Z,MEMBER,"I think `utils.fetch_timeline()` grows a new argument, `since_key`.","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",590666760,--since feature can be confused by retweets,
https://github.com/dogsheep/twitter-to-sqlite/issues/39#issuecomment-606824992,https://api.github.com/repos/dogsheep/twitter-to-sqlite/issues/39,606824992,MDEyOklzc3VlQ29tbWVudDYwNjgyNDk5Mg==,9599,simonw,2020-03-31T19:24:23Z,2020-03-31T19:24:23Z,MEMBER,"The `--since` option is actually used by four commands:

* `user-timeline`
* `home-timeline`
* `mentions-timeline`
* `search`

All of them use the same `fetch_timeline()` utility function under the hood. I should move the logic that looks up the last `since_id` into that shared function.

Question: should I have a table for each of those four methods or a single table that is used by them all? I'm leaning towards four separate tables.","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",590666760,--since feature can be confused by retweets,
https://github.com/dogsheep/twitter-to-sqlite/issues/39#issuecomment-606309165,https://api.github.com/repos/dogsheep/twitter-to-sqlite/issues/39,606309165,MDEyOklzc3VlQ29tbWVudDYwNjMwOTE2NQ==,9599,simonw,2020-03-30T23:41:31Z,2020-03-30T23:41:31Z,MEMBER,I like the separate `user_timeline_since` table solution.,"{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",590666760,--since feature can be confused by retweets,
https://github.com/dogsheep/twitter-to-sqlite/issues/39#issuecomment-606305701,https://api.github.com/repos/dogsheep/twitter-to-sqlite/issues/39,606305701,MDEyOklzc3VlQ29tbWVudDYwNjMwNTcwMQ==,9599,simonw,2020-03-30T23:30:27Z,2020-03-30T23:30:27Z,MEMBER,A better alternative would be to maintain a separate table with the last seen since value for when we ran `user-timeline` for any specific user.,"{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",590666760,--since feature can be confused by retweets,
https://github.com/dogsheep/twitter-to-sqlite/issues/39#issuecomment-606304837,https://api.github.com/repos/dogsheep/twitter-to-sqlite/issues/39,606304837,MDEyOklzc3VlQ29tbWVudDYwNjMwNDgzNw==,9599,simonw,2020-03-30T23:27:50Z,2020-03-30T23:29:31Z,MEMBER,"One option would be something like this:
```sql
select max(id) from tweets
where user = ?
and not exists (select id from tweets where retweeted_status = id)
and not exists (select id from tweets where quoted_status = id)
and not exists (select id from tweets where in_reply_to_status_id = id)
```
Might be a good idea to index those columns (after confirming that doing so would indeed speed up the query).","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",590666760,--since feature can be confused by retweets,