Option to NOT save options to ini file? #152
Comments
|
Thanks, I agree. I do think it is useful to be able to customize your defaults, but this should only happen when the user explicitly requests a change. Please see latest code, which implements your suggest that default files are only created if the user explicitly generates one ("-g y"). I think expecting code to be sane when it must deal with DICOM is setting unrealistic expectations. Before the next release we can think of about whether we really want to stick with "-g" for this command. I also think there should be three behaviors: 1) [default] read defaults but do not write them, 2) read results and write any modified results, 3) reset default file to baseline dcm2niix values. In theory, I like the idea of defaults files, as it lets the user set up the conversion to always create uncompressed (.nii.gz) or uncompressed (.nii) files without always having to explicitly write "-z y" and "-z n". |
|
I agree, the defaults are still useful. For behavior #3 perhaps you could
accept an argument -gd or -gr (similar to adding the -ba option to -b) that
would reset the defaults files?
…-Keith
-Keith
On Fri, Jan 12, 2018 at 2:23 AM, Chris Rorden ***@***.***> wrote:
Thanks, I agree. I do think it is useful to be able to customize your
defaults, but this should only happen when the user explicitly requests a
change. Please see latest code, which implements your suggest that default
files are only created if the user explicitly generates one ("-g y"). I
think expecting code to be sane when it must deal with DICOM is setting
unrealistic expectations. Before the next release we can think of about
whether we really want to stick with "-g" for this command. I also think
there should be three behaviors: 1) [default] read defaults but do not
write them, 2) read results and write any modified results, 3) reset
default file to baseline dcm2niix values. In theory, I like the idea of
defaults files, as it lets the user set up the conversion to always create
uncompressed (.nii.gz) or uncompressed (.nii) files without always having
to explicitly write "-z y" and "-z n".
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#152 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AIO8l6aORZkUQy8EcmNrat4cprS8COnCks5tJwhrgaJpZM4Rbb0S>
.
|
|
@kjamison can you test out the latest code. The options are:
If you are happy with this, please close this issue. |
|
The "-g y" and "-g n" (and default) work, but "-g o" doesn't seem to do
anything. The output says "Defaults reset", but it doesn't appear to
actually modify the .dcm2nii.ini if it exists, or create one if it doesn't.
…-Keith
-Keith
On Tue, Jan 16, 2018 at 12:05 PM, Chris Rorden ***@***.***> wrote:
@kjamison <https://github.com/kjamison> can you test out the latest code.
The options are:
-g : generate defaults file (y/n/o [o=only: reset and write defaults], default n)
1. The default behavior is -g n where the defaults are read from disk
but not saved.
2. Option -g y where the defaults are both read from and written to
disk.
3. Option -g o where the defaults are only written to disk - this will
effectively reset defaults, with any changes in the current call. For
example, dcm2niix -g o -f %s_%p will generate a defaults file where
all the values are reset except the default output filename will be the
series number and protocol name.
If you are happy with this, please close this issue.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#152 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AIO8l61Zd-KlNH3q_EdCnBMXaNYK3REjks5tLNbegaJpZM4Rbb0S>
.
|
|
Did you rebuild from the source code (instead of using an earlier release)? What operating system did you test on? When you run dcm2niix without any parameters does it report that the o option is supported |
|
Ah I was trying to run just "dcm2niix -g o" without additional inputs and
getting "Defaults reset" but no "Saving default file...."
If I include actual dicom inputs, it works as expected.
…-Keith
-Keith
On Wed, Jan 17, 2018 at 1:20 PM, Chris Rorden ***@***.***> wrote:
Did you rebuild from the source code (instead of using an earlier
release)? What operating system did you test on? When you run dcm2niix
without any parameters does it report that the o option is supported -g :
generate defaults file (y/n/o [o=only: reset and write defaults], default n)?
Below is an example I ran on MacOS where it reports both "Defaults reset"
and "Saving defaults..."
$ rm ~/.dcm2nii.ini
$ ./dcm2niix -g o ~/tst/vx
Chris Rorden's dcm2niiX version v1.0.20171215 GCC6.1.0 (64-bit MacOS)
Defaults reset
Found 1 DICOM image(s)
Convert 1 DICOM as /Users/rorden/tst/vx/vx_ep2d_diff_MDDW_20170620161913_12b (384x384x27x1)
Conversion required 0.048271 seconds (0.040404 for core code).
Saving defaults file /Users/rorden/.dcm2nii.ini
$ ls -l ~/.dcm2nii.ini
-rw-r--r-- 1 rorden staff 37 Jan 17 13:14 /Users/rorden/.dcm2nii.ini
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#152 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AIO8l8FoUB4vGKpm3sp126w82lHgZy3Hks5tLjncgaJpZM4Rbb0S>
.
|
* commit 'v1.0.20171215-8-g82b9210': (35 commits) Save defaults even if no input images rordenlab#152 Option to read, read&write or just write options rordenlab#152 Removing compiler flag to ignore deprecation of "register" in C++ Removing defunct register keyword from C++ code Argument -g generates default files only when requested rordenlab#152 fixed one of the links, and added NeuroDebian pointer Faster conversion when opts.numSeries != 0 Check if dcm2niix is latest stable release ("dcm2niix -u", Unix only) Update version notes Support implicitVR with two positions per image rordenlab#144 mod: change new line characters to _ in the dcmStr function Detect if PAR DTI volumes are sequential and slice order is consistent rordenlab#144 Detect if PAR volumes are sequential and slice order is consistent rordenlab#144 Typo Typo New command line option allowing user to extract specific data series by number. Release new version (v1.0.20171204) Support CLANG/LLVM and XCode conventions (e.g. classes only in *.hpp not *.h) Restores non-debugging compiler flags Converts TVolumeDiffusion from a class to a struct ...
|
I am late to the party here, but I don't expect command-line utilities to have persistent preferences affecting how they process data. To me, predictability is key when scripting. Currently, it appears that the only easy way to get predictable behavior, "-g o", also destroys anything that had been set by the user, which isn't very friendly. Could you consider an additional mode, that simply ignores the preference file, but doesn't overwrite it? Disclaimer: I have rarely interacted with dicoms, so I am mostly imagining how I would want it to work. |
|
@coalsont this is a good suggestion. I have added the I can appreciate your personal preference to not have a preferences file, though it is common in the field (e.g. ~/.fslconf/fsl.sh). There are two things you may want to consider:
|
* tag 'v1.0.20180614': If integer scaling is used, append " isN" to NIfTI header (rordenlab#198) Check yaml-cpp version before build. Reseting defaults does not ignore prior arguments in call rordenlab@2cd185f Add '-g i' option (rordenlab#152) Refine lossless 16-bit scaling (rordenlab#198) integer scaling option (rordenlab#198)
I think it would be a good idea if dcm2niix did NOT save options to the .ini file every time it runs, since this means the "default" options aren't really defaults in the sense of expected, but rather some unpredictable byproduct of execution history. This is particularly problematic with the "filename" option, where I would really like to have the default filename be something based on dicom parameters (ie: "%f_%p_%t_%s", with the option of using something else from time to time. However, if I ever hardcode an output name by dcm2niix -f "thisfile", all future calls to this command will produce "thisfile.nii" or "thisfilea.nii", etc...
A more sane approach would be to just not save to the ini, but maybe have an optional argument if you DO want to save parameters for future calls.
Attached is a hack/patch that I used to implement this. It's not using the same conventions you guys are using for arguments, but it at least illustrates what I mean. It won't save anything to the .ini file unless you add "--saveini y" to your dcm2niix call.
option_savenii.txt
The text was updated successfully, but these errors were encountered: