Closed wwood closed 1 year ago
Well that is much easier than I was thinking. I couldn't work out how to use the internal clusterer within the enum. But accessing the methods directly makes sense. I'll see if I can fix the tests.
Is there a reason that parse_percentage returns a decimal? https://github.com/wwood/galah/blob/31a6edd14e3417ba4cb69744fb8402f1e414a23a/src/cluster_argument_parsing.rs#L1015 Seems odd to do that and then multiply by 100 when we need it...
I guess it's just that the decimal is a bit human. Also worried a change might break coverm etc
-------------- Ben Woodcroft Group leader, Centre for Microbiome Research, QUT
From: Samuel Aroney @.> Sent: Wednesday, August 30, 2023 8:55:11 AM To: wwood/galah @.> Cc: Ben J Woodcroft @.>; Author @.> Subject: Re: [wwood/galah] Refactor clusterer choice (PR #34)
Is there a reason that parse_percentage returns a decimal? https://github.com/wwood/galah/blob/31a6edd14e3417ba4cb69744fb8402f1e414a23a/src/cluster_argument_parsing.rs#L1015 Seems odd to do that and then multiply by 100 when we need it...
― Reply to this email directly, view it on GitHubhttps://github.com/wwood/galah/pull/34#issuecomment-1698248299, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AAADX5DZAMMTP7GF3AT3DI3XXZXM7ANCNFSM6AAAAAA4CQKBHI. You are receiving this because you authored the thread.Message ID: @.***>
superseded by https://github.com/wwood/galah/pull/33
Thoughts about the last commit @AroneyS ? I actually didn't think hard about this very much - copilot did most of the coding, but seems pretty straightforward.
The tests don't pass, I guess because of the 100 issue you mentioned. If you like this can you take it to completion please?
Thanks, ben