[WIP] next release (3.0.0)

[guolinke]

This is to continue #3071, we have several new items that are not finished.
I plan to release the 3.0.0 by the end of Sep.

PRs:

• [R-package] added support for first_metric_only (fixes #2368) #2912
• Fix add features  #2754
• Add CTR conversion for categorical features (with bundling support) #2463
• Less restrictive monotone constraints #2305 (last remained part)

New Features

• GPU support on ppc64le #2937
• Unify out results of LGBM_BoosterDumpModel and LGBM_BoosterSaveModel #2604
• Add some features to R-package that are already in C API and Python package (Add support for saving weight values of a node in the R-package #2281, [R-package] exclude training data from being checked for early stopping #2472, Refit method for R-package #2369, [R-package] Add support for non-ASCII feature names #2983)
• CRAN release of course [R-Package] CRAN issues #629

Fixes (may is not included in 3.0.0):

• All known GPU bugs (lightgbm.basic.LightGBMError: Bug in GPU histogram! split 11937: 12, smaller_leaf: 10245, larger_leaf: 1704 #2793, [gpu] /root/repo/LightGBM/compute/include/boost/compute/utility/wait_list.hpp:166: void boost::compute::wait_list::wait() const: Assertion `clWaitForEvents(size(), get_event_ptr()) == 0' failed. Aborted (core dumped) #2648, OpenCL Bus error #2550, GPU build fails for lightgbm target #2126, Training on GPU fails (OSError: exception: access violation) #1717, BUG in GPU histogram #1003)
• MMLSpark issues([mmlspark] Memory Corruption #2392, JVM crash caused by LGBM_DatasetCreateFromCSRSpark #2363)

[guolinke]

cc @shiyu1994 for CTR features (and symmetric trees).

[jameslamb]

@guolinke sounds good! I think that to get to CRAN, I really need the help of you and other C++ maintainers on #3187. I'm trying to get better at C++ but it's beyond my ability to fix that issue by September.

I also think we should attempt a submission to CRAN right now (of 3.0.0-1). I think it's very unlikely that it will be accepted without #3187, but through that submission we might learn of other things that need to be fixed.

See this comment: #3187 (comment)

[guolinke]

@jameslamb no problem. sorry for missing #3187, I will try to submit it to cran.
@shiyu1994 can you help with the 32-bit build problem?

[guolinke]

@jameslamb
The submit results:

[shiyu1994]

@jameslamb no problem. sorry for missing #3187, I will try to submit it to cran.
@shiyu1994 can you help with the 32-bit build problem?

Ok.

[StrikerRUS]

@guolinke I believe RC candidates are for beta testing features that will be in the corresponding ordinary release. No new features are allowed after RC is released, only bug fixes reported during beta testing period. All features you've mentioned in your staring comment should be included in 3.1.0 version as they are non-breaking ones.

ping @henry0312 to confirm

[jameslamb]

@jameslamb
The submit results:

woo, thanks so much! You will get an email some time in the next 48 hours with requested changes, and links to logs.

You could share those on #629 or in Slack, so we don't clutter this issue.

[aakarshg]

hello, can we also add #3306 to the 3.0.0 release?

[StrikerRUS]

@aakarshg

hello, can we also add #3306 to the 3.0.0 release?

Hi!
I'm afraid that adding new feature will require one more round of testing, i.e RC2. I believe it'll be better to release 3.0.1 or 3.1.0 ASAP with merged all PRs proposing new features which all currently open and (almost) approved.

[StrikerRUS]

@guolinke Please share your thoughts about #3293 (comment) and #3293 (comment)

[guolinke]

@StrikerRUS
My purpose to release the 3.0.0rc1, is to have a time buffer to include some on-going new features into 3.0.0 .
If the next version is 3.1.0, I should release the 3.0.0 directly..

[jameslamb]

I agree, the next release should be 3.0.0 and we should do it ASAP. That was the purpose of having a release candidate.

[StrikerRUS]

@guolinke

My purpose to release the 3.0.0rc1, is to have a time buffer to include some on-going new features into 3.0.0 .

I thought that 3.0.0rc1 and 3.0.0 should be identical in terms of functionality. But I might be wrong.

[StrikerRUS]

It looks like early adopters of LightGBM have been using code from master branch and ordinary users are not aware of RC and how to install it. Given that, I think we can release 3.0.0 next week or week after week. WDYT guys?

[jameslamb]

I agree with @StrikerRUS . Especially since CRAN maintainers will be back on Monday 😀