Conversation
|
The diffs of DefaultOpTypeMatchingService and DefaultOpTransformerService got polluted by formatting and might be a bit hard to read, sorry about that. I should be using the eclipse preferences from imagej/imagej - are those the right ones? |
You and @gselzer should just stop autoformatting on save. It creates tons of problems. We can try to fix the autoformatters but it is low priority. If you want to format a code block that you just wrote, select it and then shift+ctrl+F it. It is possible to split these muddy commits using |
|
Yes, I usually turn off save actions when working on already existing files and only use them for new files, forgot it here. I know about Shift+Ctrl+F, I mostly use save actions for clean-ups (making variables final etc.). It's a pity that there's no Shift+Ctrl+F equivalent for that. Sounds good, thanks! |
For Clean Up, I usually type (on macOS): option+command+S and then C,C,return. Yeah, no official shortcut, but it's still quite fast. Unfortunately, I believe Clean Up is always global to the file, rather than operating on the current selection, so it will not help you here. This is why you see me adding |
90951d1 to
09ad613
Compare
|
Cleaned up the diffs some more, the PR should be ready for review now. |
Trims down the interface to a collection of Ops that can be listed and queried by name.
This was an artifact of the old imagej-ops that does not hold true in the new scijava-ops implementation.
This also removes the notion of a "canonical" Op name. From now on, all names that are specified for any given Op are treated equally. Note that this commit also fixes #20. It is however questionable whether the fixed behavior is desired at all. Might be adjusted in a future commit.
From OpTypeMatchingService and OpTransformerService, and rename them accordingly. This commit also separates indexing transformers from matching them by moving the indexing logic to OpService. Co-authored-by: Curtis Rueden <ctrueden@wisc.edu>
09ad613 to
9eae5fa
Compare
|
Darn, I completely forgot about this PR, and since then work continued on the migrate-imagej-ops branch, with @gselzer redoing some of the work you had already done here. I have now reconciled the branches, fixing conflicts, etc. Some of the work here might was lost: in particular, the separation of the op index into its own Anyway, all commits compile with passing tests, and are still a step forward, so I'll merge it now. |
This PR comprises a set of rather loosely related changes. The main topics are:
Feel free to approve or reject these changes separately.
The PR does not yet deal with the removal of all parts of the plugin system.
OpServiceandTypeServicestill exist and use other services internally. In particular,OpServicestill relies onPluginServiceto collect Ops and Transformers.