RLS: 0.18.0 schedule

[jreback]

so I am thinking of a RC1, friday feb 12th & release on or about friday feb 26th assuming nothing major comes up.

any objections

@jorisvandenbossche @shoyer @wesm @cpcloud @TomAugspurger @sinhrks

[wesm]

This seems fine to me.

Side issue: pandas is in need of an issue triage. We should probably move at least 500 issues to a feature deadpool to discuss and close. If we agree on a standard protocol for the triage, I can take on 50-100 personally.

For the triage, we should have 2 kinds of issues:

1. This belongs in pandas, but needs a champion to pick it up and do it (wishlist for motivated contributors)
2. This probably does not belong in pandas, and should be closed

Separately: I would also like to split out a category for issues that can be resolved as part of the libpandas / pandas 1.0 experiment. I realistically am not going to have time to start moving the ball forward there until sometime in March. I'm still scoping all the moving pieces; if anyone has time to help with this scoping effort (and isolating pieces -- like Panel and friends -- that are heading for deprecation in 0.19.x) I would really appreciate it. As soon as we scope out the work to be done and all the moving pieces it will be relatively straightforward to start making linear progress on this, with a target completion date end of 2016 / early 2017.

[jreback]

@wesm sounds all good to me.

It might make sense to create a couple of milestones for these (or labels ok too): internal refactor / triage

I find milestones a bit 'more obvious' for grouping things like this.

I also created 0.18.1 and 0.19.0 milestones.

I will (but haven't yet) created a deprecation issue of > 2 ndim.

[jreback]

release candidate is tagged: https://github.com/pydata/pandas/releases/tag/v0.18.0rc1

e-mail going out in a little

[jreback]

ok, was busy last week, but hoping to merge remaining PR's for 0.18.0 in next day or 2. Then going to do a v0.18.0rc2. Give that a week to bake then release.

[jreback]

ok ann for 0.18.0rc2 is sent here. and builds are up on conda:

conda install pandas=v0.18.0rc2 -c pandas

unless major breaks will plan for a release early next week.

[jorisvandenbossche]

@jreback Thanks!

I want already note that I won't be able to build and upload the docs this time for 0.18.0 (I am finishing my phd the coming month, so my activity level will be lower anyhow :-)). You already cleaned up a lot, but there is still an issue with the accessor API docs.

Further, just to check:

• do we have confirmation of some major pandas-relying projects (seaborn, statsmodels, ..) their tests run fine?
• @jreback did you do a performance check (full suite since 0.17.1)? If so, maybe just post it here as well to record that? (Eg an update to http://pydata.github.io/pandas/# like you did for 0.17 would be nice.)

[jreback]

oh didn't realize you couldn't build the docs - ok I can do that - are instructions in the contributing up to date? (eg for pdf and latex) do u do anything special?

not going to hold things up for perf ATM
can be addressed in next release

[TomAugspurger]

I can build the docs tonight. I’ve got latex already setup.

On Mar 9, 2016, at 5:41 PM, Jeff Reback notifications@github.com wrote:

oh didn't realize you couldn't build the docs - ok I can do that - are instructions in the contributing up to date? (eg for pdf and latex) do u do anything special?

not going to hold things up for perf ATM
can be addressed in next release

—
Reply to this email directly or view it on GitHub #12245 (comment).

[jreback]

@TomAugspurger thanks! I usually push the version update tag commit first
so u need to tag it locally and should be good
if u send me a public key ssh (email) I will setup so u can login to the server to deploy them (or I can do it)

[jorisvandenbossche]

oh didn't realize you couldn't build the docs - ok I can do that - are instructions in the contributing up to date? (eg for pdf and latex) do u do anything special?

They should be perfectly be up to date! https://github.com/pydata/pandas/wiki/Building-and-uploading-the-official-documentation (apart from the 'setting up your environment' section, some new packages are missing, statsmodels and xarray for example)

@TomAugspurger It's for next week I think, for the final release

[jreback]

@jorisvandenbossche great!

actually the requirements are up to date
(I fixed the build in the last commit)

you have to

conda remove pandas --force to avoid removing stats models / xarray

[TomAugspurger]

Was able to build the PDF, with one change in #12586

I do have some strange style stuff for the codeblocks. They're all surrounded by boxes, and occasionally (not always) have a bunch of backslashes separating the input and output.

I had the same thing happen on 0.17.1 so presumably it's my LaTeX environment. Will keep working on it.