Hello Ruy - sending you a followup off list to your email.

- rob

On May 3, 2018, at 3:16 PM, Ricardo Valle <ruyvallem@gmail.com> wrote:

        External Email - Use Caution        


Thank you for your help Rob. Could you walk me through how you ran setup_configure and configure on Mac?

I ran (from the root of the tree):
  • ./setup_configure
  • ./configure F77=/usr/local/bin/gfortran-4.9 CC=/usr/local/bin/gcc-4.9 CXX=/usr/local/bin/g++-4.9 --with-pkgs-dir=/Users/ruyvalle/Documents/fs-deps/ --disable-Werror --prefix=$HOME/fsdev_install --disable-kww-apps
  • make clean
  • make -r
Importantly, make -r said:
"Making all in trc
make[2]: Nothing to be done for `all’.”

I’ve tried using --disable-GUI-build instead of --disable-kww-apps with configure before as well, as instructed here.

I also tried:
  • cd trc
  • make dmri_train
The end of make dmri_train’s output is:
ld: entry point (_main) undefined. for architecture x86_64
collect2: error: ld returned 1 exit status
make: *** [dmri_train] Error 1

I believe the --disable-kww-apps causes the Makefile in freesurfer/trc to comment out the instructions that would make it build anything. For instance:
#bin_PROGRAMS = dmri_spline$(EXEEXT) \
# dmri_trk2trk$(EXEEXT) \
# dmri_vox2vox$(EXEEXT) \
# dmri_train$(EXEEXT) \
# dmri_paths$(EXEEXT) \
# dmri_pathstats$(EXEEXT) \
# dmri_mergepaths$(EXEEXT) \
# dmri_motion$(EXEEXT) \
# dmri_group$(EXEEXT) \
# dmri_forrest$(EXEEXT)

I tried uncommenting all the dmri_train settings in the Makefile and running make dmri_train again, and get the following:
In file included from blood.h:26:0,
                 from dmri_train.cxx:26:
vial.h:36:17: fatal error: mpi.h: No such file or directory
 #include <mpi.h>
                 ^
compilation terminated.
make: *** [dmri_train.o] Error 1

My guess is that something is not happening properly during ./setup_configure or ./configure.

By the way, do you know what the am_* options are in the Makefile?

Ruy
On May 3, 2018, at 1:35 PM, Dicamillo, Robert <RDICAMILLO@mgh.harvard.edu> wrote:

Hello Ruy,

The -r flag to will cause the build to skip some dependencies, e.g., sometimes makefiles
will rebuild the world when you don’t need them to, so if you are building only 1
file, then it can be useful.

It works for me to cd to  ./freesurfer/trc and do the same thing without using
-r, i.e.,

$ cd ./freesurfer/trc
$ rm -f dmri_train dmri_train.o
$ make dmri_train

But I’m doing this in a freesurfer tree where I have already built everything once, or I’ve
run the preconfigure, configure steps, and done a make from the top of the tree (./freesurfer).

If you see a Makefile in a directory where you want to build, then you can look for
the program name in the Makefile.  In this case, if you search for dmi_train in
./freesurfer/trc/Makefile you will see at line 819,

dmri_train$(EXEEXT): $(dmri_train_OBJECTS) $(dmri_train_DEPENDENCIES) $(EXTRA_dmri_train_DEPENDENCIES)

This is the rule to build dmri_train, or the target to build (dmri_train$(EXEEXT)) is on the left hand side of a colon,
its dependencies are on the right hand side of the colon, and the commands to build it are what follows underneath.

The variable $(EXEEXT) gets set to be whatever the executable extension is for the OS you are building on; in the
case of linux this is set to an empty string, so for linux the rule could be written as,

dmri_train: $(dmri_train_OBJECTS) $(dmri_train_DEPENDENCIES) $(EXTRA_dmri_train_DEPENDENCIES)

So you can expect that telling make to build the target “dmri_train” should do something, i.e., “make dmri_train”.

But the dependencies for dmri_train include some libraries, which do not get built just by typing
“make dmri_train” under ./freesurfer/trc when nothing else has been built in the tre.  We can look at trying
to remedy this, but I think the expectation has been that people build from the top of the tree
(or at least do that once).

- rob


On May 2, 2018, at 7:16 PM, Ricardo Valle <ruyvallem@gmail.com> wrote:

        External Email - Use Caution        


Hello Rob,

No, I mean the dmri_train binary that is part of Tracula. The main source file is ./freesurfer/trc/dmri_train.cxx. I don’t mind building the whole tree if necessary, but I haven’t been able to do so either. Do you know what the -r flag to make does?

Thank you,
Ruy
On May 2, 2018, at 6:51 PM, Dicamillo, Robert <RDICAMILLO@mgh.harvard.edu> wrote:

Hello Ruy,

Do you mean the mri_train binary that gets built under ./freesurfer/mri_train ?
(I see a ./freesurfer/dmri_poistats, ./freesurfer/dmri_tensroreig but not a ./freesurfer/dmri_train).

Things are not necessarily setup so you can cd to ./freesurfer/<subdir> and just build
what is there in isolation from the rest of the tree.

But for ./freesiurfer/mri_train, you could try the following, which worked for me on a Mac, e.g.,
after changing mri_train.c,

$ cd ./freesurfer/mri_train
$ rm -f mri_train mri_train.o
$ make -r mri_train

- rob




On May 2, 2018, at 5:37 PM, Ricardo Valle <ruyvallem@gmail.com> wrote:

        External Email - Use Caution        


Hello,

I need help recompiling dmri_train for Mac. It seems like the instructions in the Mac Developer’s don’t work.

When I cd to the trc subdirectory and run make, I am told there is nothing to do. I tried editing the Makefile, adding bin_PROGRAMS = dmri_train, at which point I got a different error. I then uncommented some other lines in the Makefile and managed to start compiling dmri_train, but I’ve been getting a never-ending sequence of errors, essentially telling me that the compiler can’t find various dependencies (boost/program_options.h, itk*.h, etc.). I’ve tried running ./configure with various flags that disable or enable various build options, copying over .h files from other locations, adding -I <path> directives for the compiler in the Makefile, etc. etc. to no avail.

Thank you,
Ruy
On Apr 30, 2018, at 5:19 PM, Ricardo Valle <ruyvallem@gmail.com> wrote:

Dear Freesurfer community,

I am trying to recompile Freesurfer in order to get a binary of the latest version of dmri_train. I have been following the instructions in the Freesurfer Mac Developer’s page and am stuck at the following:
git annex get --metadata fstags=makeinstall .

I am attaching the output (stdout and stderr) from this command to this email. Please note that I called the git-annex repository mgh rather than datasrc. The purpose of running this is so I can run make install.

Is there a way of recompiling only dmri_train or only the dmri_* programs? If not, do you know why git-annex might be giving me problems and how to solve this?

Thank you,
Ruy
<fs-annex-get.txt>

_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.

_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.