Closed dscho closed 9 years ago
@dscho @axtimwalde @acardona Any blockers to merging this?
Well, in the quest for reproducible builds for all of Fiji, I merged this. However, as always, we can iterate again if further changes need to be made!
The problem here is that it was a little bit premature. The TrakEM2 project develops in a different time frame than Fiji, and unfortunately we had to remove some TrakEM2-specific support in the 3D Viewer in order to make the 3D Viewer builds reproducible.
However, this breaks TrakEM2 until the time that somebody in that project finds the time to make a new release (no code changes are required because I did all the hard work, multiple times even because the first two iterations were rejected without further ado).
So I fear that we're in a worse spot right now than we were before because we have a completely borked version landscape in Fiji right now.
@dscho : "without further ado" is unfair: your changes are very welcome but not when they reduce performance between 2 and 3 orders of magnitude.
Additionally to reiterate that to me the top issue related to ImgLib1 is the scripting library, which is in heavy use by users, not developers, across the world. Replacing it with a scripting library based on ImgLib2 is a project that I gave up on some time ago, even if most of the scripting library in ImgLib2 works well--it is simply not finished: one tires of the Red Queen game with APIs.
Please see my upcoming mail for details of the strategy Fiji takes with regard to TrakEM2.
@acardona wrote:
to me the top issue related to ImgLib1 is the scripting library, which is in heavy use by users, not developers, across the world.
Would having a stable release of legacy-imglib1
address the issue there?
Yes. And keeping it in the Fiji distribution.
Keeping a "stable" ImgLib1 is not up for discussion.
It is time to get rid of ImgLib1...
To support TrakEM2, we redo the triangulation support here, in an ImgLib1-agnostic way.