Re: interested in "image IO" for GSoC 2015

[Carnë Draug]

Again, please keep maintainers mailing list on CC, and write your
reply at the bottom.

On 26 March 2015 at 08:57, Daniel <address@hidden> wrote:
> On Wed, Mar 25, 2015 at 8:46 AM, Carnë Draug <address@hidden> wrote:
>>
>> Better to keep the maintainers mailing list on CC.  Also, on the Octave
>> mailing list, etiquete is to do interleaved or bottom posting which makes
>> conversations easier to follow:
>>
>>   https://en.wikipedia.org/wiki/Posting_style#Bottom-posting
>>
>> On 25 March 2015 at 03:08, Daniel <address@hidden> wrote:
>> > On Tue, Mar 24, 2015 at 6:31 PM, Carnë Draug <address@hidden>
>> > wrote:
>> >>
>> >> On 23 March 2015 at 18:45, Daniel <address@hidden> wrote:
>> >> >
>> >> > I decided to work on imglib2, if you agree for GSoC. Shall I fill in
>> >> > the
>> >> > public and private form from Octave wiki? If so, I will do it send
>> >> > you a
>> >> > draft for preview and finish the from as soon as possible.
>> >> >
>> >>
>> >> I would be happy to mentor such project, but please consider the
>> >> following:
>> >>
>> >>   1 - we will only accept applications from users that have already
>> >>   contributed to the project.  It is all good to tell us what you can
>> >>   do but you will also have to show us.  I already provided you with
>> >>   some easy projects.  You haven't submitted a single patch yet.
>> >>
>> >>   2 - for your specific project, would be better to have a proof of
>> >>   concept.  Are you sure Octave has what it needs to do it?  If it
>> >>   doesn't, then the project would turn into improving Octave's java
>> >>   interface which is a very different thing and would probably require
>> >>   a mentor other than me.
>> >>
>> >>   3 - I believe that most of ImageJ plugins and functions do not use
>> >>   imglib2, instead use the original ImageJ classes.  There may be more
>> >>   value on using that instead.  It should at least make distribution
>> >>   easier for downstream packagers since ImageJ2 seems like a pain and
>> >>   is packaged by no one.
>> >>
>> >>   4 - we expect GSoC students to take it as full-time job.  I don't
>> >> know
>> >>   the details of your PhD studies, but please be aware of what we will
>> >>   expect from you time-wise.
>> >>
>> >> Carnë
>> >
>> >
>> > Hi Carnë,
>> >
>> >     Thank you very much for your guidance.
>> >
>> >     I have been trying to see different possibilities and do some
>> > planning
>> > in my mind, have not done real work yet. (It'd be better if I could
>> > start
>> > much earlier). I hope I can get something done lately.
>> >
>> >     Just for discussion, I did a bit research on the internet just now.
>> > I
>> > think the idea of an interface for importing imagej plugin so as to
>> > naturally involving packaging/managing system is really good. I found
>> > some
>> > projects/topics analogous or in relation to this.
>> >
>> >     0. There was discussion about image output format from Octave
>> > differs
>> > from ImageJ, [1][2]. The solution to this should be either simple (a
>> > converter) or complicate (maybe rewrite some library functions)
>> >
>> >     1. Matlab accesses Java, and provides Java-compatible compiler SDK.
>> > [3][4][5]
>> >
>> >     2. on Project Kenai, "GNU OCTAVE IMAGEJ BINDINGS",[6] . According to
>> > their webpage, they started this project in 2009 with three people. I
>> > have
>> > not found any further information about their progress. Maybe it is just
>> > identical to one of the followings.
>>
>> I have seen this before.  They do this by writing temporary files and then
>> reading it on the other side.  Let us not do this.
>>
>> >     3. MIJ, A Java package for running ImageJ and Fiji within Matlab,
>> > [7].
>> >
>> >     4. Miji,[8]
>> >
>> >     5. GigHub, imagej-matlab [9],  MATLAB integration with ImageJ data
>> > structures and runtime.
>> >
>> >
>> >     To sum up, we may have two different topics here. One is that the
>> > image
>> > output format is currently not compatible with ImageJ. Maybe this is
>> > related
>> > to the one of the suggestions in your first email to me --- convert
>> > octave
>> > array such that imagej/plugin can be used. The second one is that there
>> > might do not exist a Octave JVM that can be readily used to call Java
>> > routines, even load ImageJ plugin/macro directly. In the latter case, I
>> > am
>> > curious to know how you would think the usefulness to implement such
>> > JVM.
>> > Imagine, it may be able to use imagej plugin directly --- effectively,
>> > together with the prospective packaging system, it would or it has the
>> > potential to become an "enhanced" ImageJ that can readily do any image
>> > processing / post processing/ numerical analysis. Meanwhile, more Java
>> > components would be introduced. I am not sure if this could be in line
>> > with
>> > the big picture of the development of Octave. But of course, we only add
>> > what we really need to do; some balance can be made, just towards yet
>> > another Matlab, enhanced one. Anyway, I personally think this is
>> > fantastic.
>> > My question would be, in which way it is more performance-wise and/or
>> > effort-wise efficient or more manageable to write plugins/packages,
>> > using
>> > java, macro, or C++ or .m? I have not done further investigation. I only
>> > find several java source files in Octave. I will study it further.
>>
>> macros? There's no macros in Octave.  Other than that, an Octave package
>> can have C++ and m files intermixed.  I have never written java code for
>> Octave but take a look at how the old Octave java package did it.
>>
>> Again, I am unsure we have mentors for this project, so you better have
>> something sightly working and a good plan so we get an idea of what the
>> project encompasses if you want it to be accepted.
>>
>> >     Lastly, I am fully aware of the fact that the summer project
>> > requires
>> > full-time commitment. That's not an issue for me. The nature of my
>> > current
>> > work at NYU determines my flexible working schedule, especially in the
>> > summer, I can focus on the summer project for sure. Personally, I also
>> > want
>> > to take this opportunity to push myself in programming and software
>> > design.
>> > Contributing to the open source community is a great way to go.
>> >
>> >     I will let you know my progress when I have something can show you.
>> >
>> >     Thanks,
>> >
>> >     Daniel
>> >
>> > [1] http://permalink.gmane.org/gmane.comp.java.imagej/35079
>> > [2]
>> >
>> > http://imagej.1557.x6.nabble.com/ImageJ-reads-reversed-bits-from-unsigned-16-bit-TIFF-saved-by-imwrite-from-GNU-Octave-td5010285.html
>> > [3] http://www.mathworks.com/products/matlab-compiler-sdk/index.html
>> > [4]
>> >
>> > http://blogs.mathworks.com/community/2009/07/06/calling-java-from-matlab/
>> > [5] http://www.mathworks.com/matlabcentral/newsreader/view_thread/290176
>> > [6] https://kenai.com/projects/octave-imagej
>> > [7] http://bigwww.epfl.ch/sage/soft/mij/
>> > [8] http://fiji.sc/Miji
>> > [9] https://github.com/imagej/imagej-matlab.
>> >
>
> Morning,
>
> I did some more research about with octave. There are some issues that I did
> not expect. Also I am afraid it is difficult to find an available mentor in
> that area.
>
> I am generally interested in the suggested project ImageIO, and if possible,
> if no else take it, I am also interested in the N-d image processing. I
> would like to start with you specifically. But I am sorry it might be late
> to prove myself given the limited time recently.

If you don't even submit a patch, then we you will not be selected.  The
application deadline is tomorrow.  We then have until 13 of April to rank
and select students.

> I just want to avoid any
> conflict, I would like to write the template application before the deadline
> about the ImageIO project, (or also N-d image processing project), if so far
> no other student works on it.

As far as I am aware, no one else is applying for those projects.  But
please, do not think that there's a slot for every project, or that we
will select a student just because there's no one else.

> And I will also reply this email regarding to
> this, in the list, after your confirmation.

Carnë