Migrating BlenderPhotonics to pyiso2mesh and pmmc

[fangq]

• update pyiso2mesh to pick up fix surf2vol error when input face havs 5 columns pyiso2mesh#2
• update jdata to avoid importing tkinter at initial import
• add sparse_numba as dependency for pmmc (windows only)

[fangq]

@Knight680, I have committed the following two changes to fix the two problems you reported earlier

NeuroJSON/pyiso2mesh@845d0c8
NeuroJSON/pyjdata@de5b0a0

please update your iso2mesh to 0.5.1 and jdata to 0.8.1 to re-test

[Knight680]

I push the code to a new branch: PBlenderPhotonics. This branch is brandnew without submit history from master.
sparse_numba now handled by BlenderPhotonics but will show a warning from UMFPACK libraries (sparese_numba's issue)

Known issues:

1. the missing return bug in the loadmat function in jdata has been fixed on github, but not yet pushed to PyPI.
2. The loadvolume function currently only supports nii files and mat files. The batch add optics parameter function only supports json files or mat files.

[fangq]

I push the code to a new branch: PBlenderPhotonics. This branch is brandnew without submit history from master.

just to confirm this - even though this big update has no incremental commits (which is not the best practice), the updated code was still modified based upon the previous version of bp, is this correct?

unless the new code is entirely written from scratch, I still feel that committing it to the official blenderphotonics repo is still the best way to manage these versions

[Knight680]

I'm not very familiar with git, so I may have misunderstood something. Please feel free to let me know if I've misunderstood.

I think you mean commit PBlenderPhotonics to NeuroJSON/BlenderPhotonics.I did commit it, and now you can see a new, PBlenderPhotonics under the branch for this library.Since this update is a major update, if I had committed directly to the master branch there would have been a lot of conflicts (yesterday, I creat a branch from master and push the code to that, which show a lot of conflicts. Today, I delet that branch and re-create it without submit history from master), so PBlenderPhotonics is a brand new branch with no commit history from the master branch.

[fangq]

creating a new branch, committing the new code to the new branch, and push/create PR is fine. these were not the issue.

the issue is that when you created the new branch (https://github.com/NeuroJSON/BlenderPhotonics/tree/PBlenderPhotonics), the branch did not fork from the HEAD (latest commit) from the master branch. You can see from the branch link above, it is 62 commits behind the master. This means you forked from an really old commit instead of the latest commit.

because of this old fork, most of your units have conflicts and your PR can't not merge them.

to solve this, you have two ways

1. discard the PBlenderPhotonics branch
2. git checkout master # switch to master
3. git pull # update to the latest version
4. git checkout -b purepython fork master to a new branch
5. git log and make sure the top commit shows origin/HEAD
6. copy your new files to this branch, git add newly added files, and make a commit
7. git push -u origin purepython push to origin/purepython branch
8. create a new PR from the purepython branch on github

[Knight680]

Please check the updated branch: purepython.
https://github.com/NeuroJSON/BlenderPhotonics/tree/purepython
I add some data example to README for user who need load old data from Matlab/Octave workflow.

[fangq]

thanks. I see a number of files were removed in your commits, see master...purepython

can you confirm if these are intentional and these files are no longer needed?

[Knight680]

Most of removed files are matlab/octave scripts or install script. So yes, I think it's not needed for pure python version. niifile.py and nii2mesh.py now be replaced by nii2tet.py and nii2voxel.py. blender2mesh.py is replaced by blender2tet.py and blender2voxel.py.

[fangq]

niifile.py and nii2mesh.py now be replaced by nii2tet.py and nii2voxel.py. blender2mesh.py is replaced by blender2tet.py and blender2voxel.py.

if these were replaced, does that mean the new file are entirely new and unrelated to the previous files?

if they were actually renamed and modified after rename, the best way to do this is to use

git mv niifile.py nii2tet.py
git mv nii2mesh.py nii2voxel.py

commit this renamed file, then copy/overwrite both with the new files. this way, the history is still kept for these units.

can you confirm if the nii2tet and nii2voxel file were modified from the older ones?

[Knight680]

Well, I prefer to considet the nii2tet and nii2voxel are entirely new althought they present a silimar function.

old file:

• niifile.py: the file select box
• nii2mesh.py: call v2s to convert the voxel data to tet mesh

new file:

• nii2tet.py: including file select, data load and v2s fucntion for pmmx workflow
• nii2voxel.py: including file select, data load and pre-process voxel data for pmcx workflow

So, both nii2tet and nii2voxel include old niifile and nii2mesh function. But if you want to keep the old files, I can renamed them.

[fangq]

So, both nii2tet and nii2voxel include old niifile and nii2mesh function. But if you want to keep the old files, I can renamed them.

based on what you said above, I suggest to use git mv to rename.

[Knight680]

Done. Please have a check and feel free to let me know if there is any other problem.

[fangq]

looks good! please create a PR against the master branch and in the title of the PR, make sure you have "fix #12"