Closed cbm755 closed 9 years ago
I would be happy for us to be an Octave package. (I understand that for Matlab users, the only change required is to add the inst/
subdirectory to the PATH.) Do you want to go ahead and reorganize the repository in an appropriate way?
I added travis-ci integration earlier this morning; it would be great if you could adjust the make test
command appropriately.
I believe the license is the 3-clause BSD license. My understanding is that we are free to modify the sources, rename the software (e.g. we might rename the repository to doctest-for-octave
), and distribute the result as long as we make sure to preserve the original copyright and license statements. However, IANAL :smiley:
I agree that it is 3-clause BSD. But also IANAL.
But that would be a fork for sure.
Names: "octave-doctest" or perhaps try to avoid both octave and matlab: "doctestm", "doctestem", "Doc Test'em" -- ala Duke Nukem :)
I like these names ;)
I recall that you had contacted Thomas Smith a little while ago. Did he get back to you at all?
I wrote to him on some contact form on BitBucket. But maybe that just goes to /dev/null.
I notice he has a github account: so ping @tgs
Here's my message:
Hi Thomas,
Thanks for this code. Are you still interested in maintaining it?
I've been making a few changes to better-support Octave. Currently, we have two octave forge packages using this for testing, with a rough idea to have the full Octave core using doctests eventually.
I've also been fixed bugs that are true on Matlab as well.
I've been organizing this around catch22's github clone but could organize it elsewhere (i.e., your bitbucket).
thanks again,
Colin
@catch22 and I'm sorry that should have said "we have been organizing".
It would be great if he's onboard with our changes!
It is very important to have this as an Octave Forge package. To have an installable and loadable package is necessary if documenation checking shall be available to more than one developer in an easy way (which is a prerequisite for using it).
Otherwise everybody would have to download it manually and we have to mess with load paths in everybody's workspace.
@oheim This discussion was purely about the organization of the repository (and not about whether doctest should be available on OctaveForge). However, we agreed on organizing the repository in the "standard" OctaveForge way.
@catch22 Okay, I just wanted to leave that comment somewhere ;-)
IMHO you have to choose a new name (clause 3), must not name T. Smith as the package author (clause 3) and keep the copying conditions (clause 1) as well as the copyright statements of course.
However, you may use GPL for new function files in the package, because the license is compatible with GPL.
@oheim Thanks, that was very helpful. So unless @tgs endorses this modified version, is the following procedure correct?
doctest-for-matlab
before we distribute it on octaveforge (as to not violate clause 3).(I want to keep the license as BSD.)
I am not familiar with the BSD-style licenses, but it's probably not needed to have a separate CONTRIBUTORS file when you have copyright statements from all contributors in COPYING.
But if the name is simple ("doctest") then I don't think we can realistically claim this in the license:
Neither the name of Doctest nor the names of its contributors may be
used to endorse or promote products derived from this software without specific
prior written permission.
E.g., given that doctest is already used in Python community.
I wonder if its still "modified BSD" if you just put:
The names of the contributors may not be
used to endorse or promote products derived from this software without specific
prior written permission.
I dislike license proliferation so probably bad idea to be rolling our own.
See PR #36 for some copyright updates.
With the simple name the product name stuff should be removed from the license. Then it will be an exact ModifiedBSD.
I don't really understand Octave pkg naming. Because things often have generic names like "image" or "interval", which is good within Octave. But then outside, should they be called the "Octave-Forge Image package"... @oheim opinion?
It's a local name within the octave forge namespace. For example Debian packages will then carry the name octave-packagename. And in any kind of documentation you have to say “Octave Forge name package” to make clear what you mean. I personally like the simple names as long as the name matches the content.
However, you should probably ask Carnë about the actual backgrounds and conventions.
Thanks @oheim. @catch22 so how about:
git.code.sf.net/p/octave/doctest
)Does that sound reasonable? We can also run it by Carnë when I ask him to setup the sourceforge stuff.
Sounds all good to me.
(I want to keep the license as BSD.)
For the record, I prefer GPLv3+ but am happy enough with Modified BSD.
Hi! I'm terribly excited that you folks are making an Octave version! I don't use Matlab or Octave at work any more, so I don't want to be in charge of anything. Fork with all my blessings!
@psexton has an active fork of this plus Matlab-xunit going, I think he's been mostly focused on making it work with non-ancient versions of Matlab, and fixing bugs. That system uses classdef, so maybe it still doesn't work in Octave?
My interpretation of that part of the license is simply that you can't advertise your version as "Thomas Smith's Doctests for Matlab!" That said, I would be happy to give you another license, eg 2-clause BSD. Let me know.
Thanks @tgs. I was not aware of the other fork, thanks for that.
I don't think another license from you is needed, but appreciate the offer.
@catch22 So do you think we should rename the github project to "octave-doctest"? We should do so before an 0.3.0 release...
Upstream Octave has preference for mercurial so you might consider switching to that. (I have no strong opinion either way).
@tgs Thanks for the kind reply!
@cbm755 That sounds good to me. Should I just go ahead and rename the repository?
Should I just go ahead and rename the repository?
Sure! There may be some fallout but I think it mostly effects @oheim and I: we'll deal with it.
Done; travis-ci is running again & I've fixed some links. Feel free to close this issue as you deem appropriate.
Thanks!
@catch22 Let's (eventually) pick one of these.
inst/
, various other rules. PR #34.we can be a continue as a Matlab package (note this is not well-defined) and have some script that generates Octave packages.[going with first approach].Also:
DESCRIPTION
file].Is there any legal problem problem here? [no]octave-doctest
.]DESCRIPTION
file to root. 4f1bc7929c4d0e867f08e63bbce9d6fc29b8ed6b