[Mrtrix-discussion] Fwd: MRtrix mrconvert in windows
Jesse Ross-Jones
jesse.rj at gmail.com
Fri Jun 14 01:27:35 PDT 2013
Hi Ben,
Thank you for the reply. Indeed compiling gtk+ from source was a nightmare.
Good to know about the issues with selfcompiled gtk, I guess I will switch
to using the binary gtk+ installer.
I am currently at the point where I can run the python build script and
properly compile the module I need. The problem I am having is just running
the compiled exe on another computer. I have tried two different methods,
one was to copy the exe and all the dlls that it was asking for to the PC,
the other was to install the binary release of MRtrix and replace mrconvert
with the mrconvert module I compiled. Both methods result in the 0xc0000142
error.
When you say "copy the compiler runtime" do you mean the mingw bin
directory? (This would mean installing mingw on the new computer and
copying the compiled gtk, gtkglext, GTKMM and GSL into mingw?)
Thank you,
Jesse
On Tue, Jun 11, 2013 at 5:54 PM, Jeurissen Ben <ben.jeurissen at ua.ac.be>wrote:
> Dear Jesse,
>
> (You are aware that there is a windows binary release on the mrtrix
> page, right?)
>
> Compiling and linking the 32-bit mrtrix command line tools on Windows is
> quite straightforward. Assuming you are using mingw as a compiler and have
> python installed, the latest version should compile out of the box using
> the python build script. That is assuming you have all the dependencies
> correctly installed (which apparently you have?).
>
> One big issue that I had is that the available windows binaries of the
> dependencies would not work together nicely. Some of the binaries date back
> to 2006, while others were built only last year. Some of them were built
> with very specific build and link flags, making them incompatible with
> other binaries... For compilation and linking, this usually was not an
> issue, but at runtime I experienced erratic numerical and multi-threading
> behavior (i.e. very different numbers coming out on windows as opposed to
> linux) and many runtime crashes (especially when using mrview). In the end,
> I used the official gtk binary package (compiling gtk from scratch is a
> nightmare), compiled gtkmm and gsl myself and compiled an unstable svn
> version of gtkglext. Only then, the resulting binaries would be stable.
>
> So, be very careful with the windows binaries you create, as being able
> to compile,link and run them does not guarantee correct operation.
>
> The exact versions I used are:
>
> GCC: 4.7.0 (thread model win32, dwarf2 exceptions)
>
> GTK+: 2.24.10 (from the all-in-one bundle on
> http://www.gtk.org/download/win32.php)
> gtkglext: latest git release (self compiled)
> GTKMM: 2.24.2 (self compiled)
> GSL: 1.15 (self compiled)
>
> With those binaries correctly installed in a msys environment with
> python, you can just run "python build" and the entire mrtrix package will
> be built.
>
> About the 0xc0000142 error, I'm pretty sure this is a dll-related issue.
> The only difference on another computer running the same operating system
> are the dlls right? What dlls have you copied over to the other machine?
> Did you copy the compiler runtime? (In my experience Dependency walker
> doesn't always show all required dlls).
>
> Cheers,
> Ben
>
> Ben Jeurissen, Ph.D.
> Post-doctoral researcher
> Vision Lab, Dept. of Physics
> University of Antwerp
> Universiteitsplein 1, N.1.18
> B-2610 Wilrijk, Belgium
> Phone: +32 3 265 24 77
> Email: ben.jeurissen at ua.ac.be
> Url: http://visielab.ua.ac.be/people/ben-jeurissen
> ------------------------------
> *From:* mrtrix-discussion-bounces at public.nitrc.org [
> mrtrix-discussion-bounces at public.nitrc.org] on behalf of Jesse Ross-Jones
> [jesse.rj at gmail.com]
> *Sent:* 11 June 2013 16:08
> *To:* mrtrix-discussion at public.nitrc.org
> *Subject:* [Mrtrix-discussion] Fwd: MRtrix mrconvert in windows
>
> Dear MRtrix community,
>
> I have compiled MRtrix in windows after a lot of work installing the gtk
> libraries. Now however, when I try to run mrconvert on another computer
> running the same operating system (Win8 32 bit) I get the error: *The
> application was unable to start correctly (0xc0000142). *At first I
> thought this was an error with DLLs, however using dependancy walker I have
> found this is not the case. I am wondering now if the error is perhaps
> there is a flag I must use during the compilation in order to use the exes
> on another machine. Looking at the build script I see the following option
> * -system=X specify the architecture to build*. Could this be
> used to compile the executable for use on another system?
> Once I am able to compile and run MRconvert in windows, I would be happy
> to contribute the process for others in the future to follow. I have
> already made a list of all the dependencies and the required versions
> needed to compile.
>
> thank you,
> Jesse
>
> ---------- Forwarded message ----------
> From: Jesse Ross-Jones <jesse.rj at gmail.com>
> Date: Mon, Jun 3, 2013 at 9:21 AM
> Subject: Re: MRtrix mrconvert in windows
> To: Jeurissen Ben <Ben.Jeurissen at ua.ac.be>
>
>
> Dear Ben,
>
> After much work I have finally succeed in compiling MRtrix in windows. It
> is quite a bit of work to find the correct versions of the dependencies.
> Now however when I try to run the compiled exe on another windows (same OS
> ) machine I get an error:
> The application was unable to start correctly (0xc0000142). And when I
> dig a bit further I find this is likely related to problems with 64 bit and
> 32bit libraries. However when I try with another 32 bit machine it also
> doesnt work.
>
> I was wondering if you would know either how to compile MRtrix such that
> the executables can be run on other machines, or perhaps there is a way to
> package the exes with the required libraries?
>
> Thank you very much
> Jesse
>
>
> On Wed, Apr 10, 2013 at 10:56 AM, Jeurissen Ben <Ben.Jeurissen at ua.ac.be>wrote:
>
>> Dear Jesse,
>>
>> I have no experience with building mrtrix using an IDE. Best place to
>> ask is on the mailinglist. If there's no answer, it probably means no-one
>> has tried or succeeded before.
>>
>> I'm pretty sure it's possible to make a debug build with the mrtrix
>> build environment and then do the debugging of the binary with your
>> debugger/IDE of choice.
>>
>> If you insist on doing everything in the IDE, then you will have to
>> look into how the IDE handles dependencies and convert the dependency
>> resolving mechanism of mrtrix to something your IDE can handle.
>>
>> Cheers,
>> Ben
>>
>> Ben Jeurissen, Ph.D.
>> Post-doctoral researcher
>> Vision Lab, Dept. of Physics
>> University of Antwerp
>> Universiteitsplein 1, N.1.18
>> B-2610 Wilrijk, Belgium
>> Phone: +32 3 265 24 77
>> Email: ben.jeurissen at ua.ac.be
>> Url: http://visielab.ua.ac.be/people/ben-jeurissen
>> ------------------------------
>> *From:* Jesse Ross-Jones [jesse.rj at gmail.com]
>> *Sent:* 08 April 2013 08:37
>>
>> *To:* Jeurissen Ben
>> *Subject:* Re: MRtrix mrconvert in windows
>>
>> Hi Ben,
>> Thank you once again for your previous help.
>> I was wondering if you might be able to help me get in contact with
>> someone who knows how to compile MRtrix in a debugging environment? I
>> posted a message on the mailing list last week but did not receive a
>> response.
>> Much appreciated and Best Regards,
>> Jesse
>>
>>
>> On Thu, Mar 21, 2013 at 2:54 PM, Jesse Ross-Jones <jesse.rj at gmail.com>wrote:
>>
>>> Hi Ben,
>>> Being unable to run the build in windows, i've started to compile the
>>> source in linux and try to extract the necesarry libraries from there,
>>> however, i see over a hundred .o libraries .... Do you think you would be
>>> able to provide an example of the -l switch you are using to compile? Would
>>> having a successful build in linux be of any help when compiling in windows?
>>> Best Regards,
>>> Jesse
>>>
>>>
>>> On Thu, Mar 21, 2013 at 10:34 AM, Jesse Ross-Jones <jesse.rj at gmail.com>wrote:
>>>
>>>> Thanks very much!
>>>> Following the build procedure i run the build but I run into the
>>>> following even though gsl is included in my path..
>>>>
>>>>
>>>> failed with output:
>>>> In file included from lib/ptr.h:26:0,
>>>> from lib/file/dicom/image.h:33,
>>>> from lib\file\dicom\image.cpp:44:
>>>> lib/mrtrix.h:69:29: fatal error: gsl/gsl_version.h: No such file or
>>>> directory
>>>> compilation terminated.
>>>>
>>>> Is there another path python is using?
>>>>
>>>> Much appreciated,
>>>> Jesse
>>>>
>>>>
>>>> On Thu, Mar 21, 2013 at 9:25 AM, Jeurissen Ben <Ben.Jeurissen at ua.ac.be>wrote:
>>>>
>>>>> This list is generated by the python build script as part of the
>>>>> build procedure. The easiest way would be to run this script and see what
>>>>> libraries it wants to link. Alternatively you can look through the build
>>>>> script and find the relevant libraries yourself. However, they are not in
>>>>> there hardcoded but I think they are obtained through the pkg-config
>>>>> package.
>>>>>
>>>>> Cheers,
>>>>> Ben
>>>>>
>>>>> Ben Jeurissen, Ph.D.
>>>>> Post-doctoral researcher
>>>>> Vision Lab, Dept. of Physics
>>>>> University of Antwerp
>>>>> Universiteitsplein 1, N.1.18
>>>>> B-2610 Wilrijk, Belgium
>>>>> Phone: +32 3 265 24 77
>>>>> Email: ben.jeurissen at ua.ac.be
>>>>> Url: http://visielab.ua.ac.be/people/ben-jeurissen
>>>>> ------------------------------
>>>>> *From:* Jesse Ross-Jones [jesse.rj at gmail.com]
>>>>> *Sent:* 21 March 2013 09:10
>>>>> *To:* Jeurissen Ben
>>>>> *Subject:* Re: MRtrix mrconvert in windows
>>>>>
>>>>> Hi Ben,
>>>>> Thank you for the quick reply. I am a little at a loss as to how to
>>>>> identify exactly which libraries i need to link to with -l. Currently none
>>>>> are linked and that would be why I am getting those errors. Is there a way
>>>>> generate some sort of list, or manual searching that needs to be done?
>>>>>
>>>>> Much appreciated!
>>>>> Jesse
>>>>>
>>>>>
>>>>> On Wed, Mar 20, 2013 at 1:11 PM, Jeurissen Ben <Ben.Jeurissen at ua.ac.be
>>>>> > wrote:
>>>>>
>>>>>> Dear Jesse,
>>>>>>
>>>>>> I don't have any recent experience with eclipse.
>>>>>>
>>>>>> If you encounter linking issues this probably means you are missing
>>>>>> library flags in your link commmand.
>>>>>>
>>>>>> You need to add the library paths with the -L switch and link the
>>>>>> specific libraries with the -l switch.
>>>>>>
>>>>>> What link related switches do you currently have?
>>>>>>
>>>>>> Cheers,
>>>>>> Ben
>>>>>>
>>>>>>
>>>>>>
>>>>>> Ben Jeurissen, Ph.D.
>>>>>> Post-doctoral researcher
>>>>>> Vision Lab, Dept. of Physics
>>>>>> University of Antwerp
>>>>>> Universiteitsplein 1, N.1.18
>>>>>> B-2610 Wilrijk, Belgium
>>>>>> Phone: +32 3 265 24 77
>>>>>> Email: ben.jeurissen at ua.ac.be
>>>>>> Url: http://visielab.ua.ac.be/people/ben-jeurissen
>>>>>> ------------------------------
>>>>>> *From:* Jesse Ross-Jones [jesse.rj at gmail.com]
>>>>>> *Sent:* 19 March 2013 16:19
>>>>>> *To:* Jeurissen Ben
>>>>>> *Subject:* MRtrix mrconvert in windows
>>>>>>
>>>>>> Hi Dr. Jeurissen,
>>>>>>
>>>>>> First of all, thank you very much for your work on the MRtrix
>>>>>> project.
>>>>>>
>>>>>> I am sending you this email in hopes that you might be able to help
>>>>>> me to compile mrconvert (part of MRtrix) in windows.
>>>>>>
>>>>>> Following the instructions:
>>>>>> http://www.brain.org.au/software/mrtrix/install/windows.html
>>>>>> I have downloaded the dependencies and source.
>>>>>>
>>>>>> I have set up a project in eclipse using the mingw compiler and I
>>>>>> have copied the mrconvert.cpp into a new project and added paths to eclipse
>>>>>> pointing to all the header files required by mrconvert and its includes.
>>>>>> However.... I am running into linking issues.
>>>>>> The errors I am receiving are all "undefined reference" for example:
>>>>>>
>>>>>> undefined reference to `Glib::Timer::elapsed() const' mrconvert
>>>>>> line 453,
>>>>>>
>>>>>> undefined reference to `gsl_isnan' mrconvert line 453
>>>>>>
>>>>>> undefined reference to `gsl_matrix_alloc' mrconvert line 153
>>>>>>
>>>>>> I would very much appreciate your insight into compiling MRtrix in
>>>>>> windows and help to compile mrconvert.
>>>>>>
>>>>>> Best Regards,
>>>>>> Jesse
>>>>>>
>>>>>
>>>>>
>>>>
>>>
>>
>
>
> _______________________________________________
> Mrtrix-discussion mailing list
> Mrtrix-discussion at www.nitrc.org
> http://www.nitrc.org/mailman/listinfo/mrtrix-discussion
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.nitrc.org/pipermail/mrtrix-discussion/attachments/20130614/484a7722/attachment-0001.html
More information about the Mrtrix-discussion
mailing list