-
Notifications
You must be signed in to change notification settings - Fork 217
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Removed curve_fit #56
Conversation
This is just a small simplification, as the curve_fit functionality was broken out into CurveFit.jl.
Guess I should have looked here before filing #57... |
OK, backed out one commit to the respective branch and this PR to keep |
As soon as |
Once CurveFit.jl is released, I'm happy to merge this. |
@cmcbride, LGTM. I've given you temporary admin access to move CurveFit.jl into JuliaOpt. |
@mlubin Didn't realize that was necessary, but I can do that no problem. Should I "transfer ownership", fork it into JuliaOpt, or just |
@cmcbride I think the best way is to move over the original repo itself. |
OK, @mlubin, it's transferred. Maybe |
Yeah, that's reasonable. Just let us know when that happens. |
Uh oh, guess we failed doing our homework: Thoughts for a new name? Also, this conversation is starting to feel OT for this issue. |
How do the packages differ in functionality? JuliaOpt/CurveFit seems more general |
@mlubin I agree JuliaOpt/CurveFit seems more general. But they appear to be different takes on the same niche, so it makes sense to just rename this package (which was conceived a bit after the other ones creation). As the author of (the first) Possibilities:
|
ModelFit sounds fine to me, but I'd be much more in favor of combining the two packages. We'll see what @pjabardo says. |
I feel like in general that the "better developed" package should get the name, its looks like from http://pkg.julialang.org/?pkg=CurveFit&ver=0.3 that it doesn't even work. But a merge would be best. |
Hopefully we can resolve this by discussion with @pjabardo before resorting to any drastic measures. I've opened an issue on pjabardo/CurveFit.jl. |
it does not look like @pjabardo is responding quickly. Perhaps we can consider a name change to ModelFit and work on merging |
Given that merging/renaming packages in METADATA is a bit of a mess right now, I think its worth trying to resolve this first, but if others would prefer to just rename it for now that's okay with me. We don't really have a policy for unresponsive package authors, though it's something that we'll have to deal in general with sooner or later. |
I was thinking of bringing it up at JuliaCon |
Sorry for the delay. I was out hiking and without Internet connection for the past two weeks. I fixed the problem on the repository three weeks ago but I experienced some trouble updating METADATA.jl, being new to git and all. I am not home but I will try to install julia (and git) on a computer today or tomorrow. One issue that I've been having is with managing the packages. I find the idea of using git/GitHub brilliant but for package writers the docs are not clear and from my experience, Pkg.publish only works if one has write access to METEDATA. It is kind of confusing. A short recipe on how to update a package on metadata.jl would be helpful. Sorry for the mess and for disappearing. Paulo Em dom, 15 de jun de 2014 03:22 BRT Cameron McBride escreveu:
|
Hi Paulo (@pjabardo) -- are you interested in merging your CurveFit.jl with the JuliaOpt/CurveFit.jl? @mlubin -- my intention was to keep the name |
Can you elaborate on what the problem with this is? Renaming a package seems to work fine as long as you don't immediately make a new, unrelated package with an old name like happened with the whole StatsBase / Stats fiasco. Now that was a mess, I have to agree. |
@StefanKarpinski How clean would it be if we published ModelFit and then eventually it was renamed and replaced CurveFit after integrating the functionality of |
It wouldn't be a problem – again, as long as you don't create a new, unrelated package with the same name, at least without letting some time elapse between (say one release cycle). |
The more I think about it, the more |
|
@lindahua fair enough. It's definitely not a likelihood estimator -- just a minimizer. Funny how "Curve" works well just because it's such a superficial word, so people know what to expect. |
FWIW, I strongly agree with @mlubin. I believe the time to try to keep things consolidated is now, as the number of julia packages will only explode over time making the situation worse. My opinion: the merger is worth pushing for. I think the question is when and how. I skimmed the codes, and it seems easy to implement a solution that will make both project directions happy. Maybe it would be helpful to consider two possibilities:
(I think these address the issues @StefanKarpinski mentioned above.) If (1) is the chosen solution, I think it makes sense to abandon the current
I'm happy with either possible solution. Someone just has to decide. @pjabardo? @mlubin? someone else? |
Hate to do this, but is |
What about |
That seems like a much better name to me. |
Yes. I think this is probably the best solution. Paulo Em Quarta-feira, 18 de Junho de 2014 20:26, Dahua Lin notifications@github.com escreveu: What about LeastSquareFit or LsqFit, which seems to accurately reflect what the package does. |
I like either of those. |
|
|
I think a rename should be fine. I'll push a change with the internal name changes. |
Done, both you and @pjabardo have commit access. |
OK, as soon as we get LsqFit.jl listed in METADATA then I think this PR is ready to apply. |
What was the final decision on this? I got a bit lost |
I'm not sure myself. |
Sorry, I've been consumed with work and a family move over the past couple weeks. This patch (or equivalently removing the functionality) should be applied, and was waiting for JuliaOpt/LsqFit to be listed in METADATA (which I never got to submitting). Someone else is welcome to submit LsqFit.jl to METADATA, and then apply this. Otherwise, I'll try and look at it sometime in a week-ish. |
@cmcbride would you please go ahead to publish LsqFit? It seems to me that there's no objection to this. |
If someone wants to grant me access to LsqFit, I'd be happy to publish it. On Sunday, July 6, 2014 at 7:59 AM, Dahua Lin wrote:
|
Done @blakejohnson |
That'd be great, @blakejohnson. Like I said, I'm still fairly consumed by a number of work/home issues (waaaay behind at work at the moment). |
Bump. I think LsqFit is available now. |
Yes, I think this can be merged. |
It needs some rebuilding to be put into a merge-able state. |
Bump again. @cmcbride, could you resolve the merge conflicts? |
OK, I'll rebase and resolve the merge issues. Sorry -- I've been pretty busy at work, but I'll have a look at it over the weekend. |
This is to support the discussion of #55 and remove the functionality that was broken out into
CurveFit
.There were two commits: one that removed
curve_fit
(and theDistributions
dependency) and a followup that removedlevenberg_marquardt
.