ENH: Update multi-stage recon-all directives #1991
Conversation
effigies
force-pushed
the
enh/reconall_directives
branch
from
5cfbea9 to
4b8314f
Compare
Codecov Report
@@ Coverage Diff @@
## master #1991 +/- ##
==========================================
- Coverage 72.15% 72.14% -0.01%
==========================================
Files 1117 1117
Lines 56456 56487 +31
Branches 8112 8125 +13
==========================================
+ Hits 40736 40755 +19
- Misses 14443 14456 +13
+ Partials 1277 1276 -1
Continue to review full report at Codecov.
|
effigies
force-pushed
the
enh/reconall_directives
branch
from
4b8314f to
1fa5adb
Compare
| if all((name == 'hippocampal_subfields_T2', | ||
| isdefined(self.inputs.hippocampal_subfields_T1) and | ||
| self.inputs.hippocampal_subfields_T1)): | ||
| trait_spec.argstr = trait_spec.argstr.replace('T2', 'T1T2') | ||
| argstr = trait_spec.argstr.replace('T2', 'T1T2') | ||
| return argstr % value |
There was a problem hiding this comment.
This part here was a (minor) bug, in that trait_spec is non-ephemeral, so changing trait_spec.argstr changes it for all future calls. Unlikely to affect anything except the doctests, but still worth cleaning up.
| @@ -950,14 +989,24 @@ def _is_resuming(self): | |||
| def _format_arg(self, name, trait_spec, value): | |||
| if name == 'T1_files': | |||
| if self._is_resuming(): | |||
| return '' | |||
| return None | |||
There was a problem hiding this comment.
Switched to returning None just because that avoids the extra spaces in the cmdline string, making the doctests less painful to debug.
effigies
changed the title
|
Marking this [WIP] until I verify that we don't need to rework the |
effigies
force-pushed
the
enh/reconall_directives
branch
from
cd3d917 to
e9ed516
Compare
|
Needed to update |
effigies
force-pushed
the
enh/reconall_directives
branch
from
6bc955a to
e9ed516
Compare
effigies
changed the title
|
This is working as expected on my end. |
| 'surf/lh.inflated.H', 'surf/rh.inflated.H', | ||
| 'surf/lh.inflated.K', 'surf/rh.inflated.K'], []), | ||
| ] | ||
| _autorecon2_lh_steps = [ |
There was a problem hiding this comment.
only a nitpick - is there a reason this is called _lh_ as opposed to say _hemi_
otherwise looks reasonable to me.
There was a problem hiding this comment.
Because I specify lh in the filenames. I then make a rh copy below, just swapping the filenames out.
effigies
force-pushed
the
enh/reconall_directives
branch
from
e9ed516 to
8fcc097
Compare
The goal of this PR is to give access to most* multi-stage flags, in particular the
autorecon2-volonlyandautorecon-hemiflags.At present, estimating thread consumption is difficult, and
recon-all -parallel -openmp Ncan consume up to 2N threads at OpenMP-enabled sections of the surface processing sections ofautorecon2andautorecon3. This change allowsautorecon2-volonlyto be tagged as consuming N threads, and (for instance) two invocations ofautorecon-hemito each be tagged as consuming N threads.This also updates the docstring to document and test the more recent commandline altering changes.
* The one that's missing is
gcatrain-iter, which takes 2 arguments and would thus be fiddlier to deal with. (I'm not really sure whygcareg-itersandgcareg-tolare included in that section; but I wouldn't call them directives.) I can put it in for completeness sake, but given that I'm not even sure how to test it, that feels like an invitation to write a bug.