Skip to content
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

[release-0.4] WIP: backports for 0.4.4 #15385

Merged
merged 58 commits into from
Mar 15, 2016
Merged

[release-0.4] WIP: backports for 0.4.4 #15385

merged 58 commits into from
Mar 15, 2016

Conversation

tkelman
Copy link
Contributor

@tkelman tkelman commented Mar 7, 2016

If you'd like to nominate anything that doesn't already have the label go ahead and leave a comment here - especially doc-only or test-only changes are usually fair game.

TODO's:

hayd and others added 30 commits March 7, 2016 04:38
When a code block included a ```, it would not be rendered correctly.

(cherry picked from commit 60eda14)
ref #14638
(cherry picked from commit c1005bf)
ref #14301
this was causing several package test failures on jb/functions

(cherry picked from commit ec3d632)
ref #14673
(cherry picked from commit a49bef9)
ref #14680
All the memory accesses are contiguous, so there's no point in copying
data around.

(cherry picked from commit 1d10910)
ref #14703
(cherry picked from commit bcc6e2f)
ref #14630
fixes #14467 performance regression (and some)

(cherry picked from commit 9dbc3ab)
ref #14667
(cherry picked from commit d0fbd71)
ref #14519
Fixes #13200

(cherry picked from commit 59ca442)
ref #14773
to a CHOLMOD.Dense

(cherry picked from commit 2b6c143)
ref #14812

Use SparseMatrix instead of SparseArrays for release-0.4
a new type's supertype was not taken into account while instantiating
the types of its fields

(cherry picked from commit 0dd74d5)
ref #14847
(cherry picked from commit a0b5414)
…4 system blas

and link cholmod to lapack, needs dpotrf

(cherry picked from commit 4ec8918)
this should only matter in the rare case where you're building reference
lapack from source against an ILP64 blas

(cherry picked from commit 06c9fb7)
(cherry picked from commit 9cda53d)
ref #14830
[ci skip]

(cherry picked from commit abdfc24)
(cherry picked from commit 3051685)
ref #14899
Fix the error message and make the dimension output format consistent with how `size(mtx)` does it.
(cherry picked from commit f197e8c)
ref #14907
(cherry picked from commit e9a16e9)
ref #14906
timholy and others added 5 commits March 7, 2016 09:04
(cherry picked from commit 27faae4)
ref #15318
fix error reporting conditions for BLAS.axpy!, BLAS.copy!,
diag and getindex of Tridiagonal and SymTridiagonal, and umfpack

(cherry picked from commit 4e207ac)
ref #14824
@nalimilan
Copy link
Member

RPM builds went fine, so +1

matthieugomez and others added 6 commits March 8, 2016 18:43
grammar fix in doc/manual/modules.rst

(cherry picked from commit f06868d)
Use versions from the build system rather than win-extras (which downloads
from opensuse)

Remove gcc dlls from downloaded nightly on appveyor

(cherry picked from commit c1e565c)
ref #15285
from juliacache to fix ABI incompatibilities and get a release out
mostly reverts e4bd799
…ks around this issue

(cherry picked from commit af81d6c)
ref #13491
@musm
Copy link
Contributor

musm commented Mar 13, 2016

can include #15407

mustafa and others added 2 commits March 13, 2016 10:57
(cherry picked from commit ae775ea)
ref #15407
@toivoh
Copy link
Contributor

toivoh commented Mar 13, 2016

#15309? I saw you added the back port pending label to it.

@tkelman
Copy link
Contributor Author

tkelman commented Mar 13, 2016

@toivoh I guess I can put that in, it'll require re-running PkgEval. My previous run at 1d27400 is showing regressions in DictFiles, ExperimentalAnalysis, and FunctionalData, so will need some more tweaking and a re-run of PkgEval. Logs at https://gist.github.com/bbaaf1b809a2b50d0d56

I bisected the FunctionalData failure to 132cb20, @yuyichao can you take a look?

@tkelman
Copy link
Contributor Author

tkelman commented Mar 13, 2016

DictFiles failure starts happening at the same commit. ExperimentalAnalysis was a PyCall segfault that I'm having trouble reproducing.

@yuyichao
Copy link
Contributor

The issue with FunctionalData is that the local function pass is now refering to a global function vect that specialized on (and therefore has a reference to) a type (_somedummytype) that is not defined on the worker process. I'm not really familiar with the serialization of functions/types (and when is @everywhere necessary) to tell what's the right fix....

Given this commit is causing too much trouble and is fixing unlikely segfault for a unsupported usage (#265) I think we can leave this out for backporting.....

tkelman and others added 3 commits March 13, 2016 16:46
They are still given to the user in macro arguments,
otherwise LineNumberNode seems to be used consistently.

(cherry picked from commit 3a16b19)
ref #15309
@toivoh
Copy link
Contributor

toivoh commented Mar 14, 2016

@tkelman: Yeah, or if it's too much trouble then you can just wait with it.

and fix some broken links
@tkelman
Copy link
Contributor Author

tkelman commented Mar 15, 2016

A handful of failures here that all seem intermittent, I can't reproduce any of them locally. https://gist.github.com/0c394be0a2f0b4e0d736

I'll merge and tag this soon, assuming my incoming trivial doc fix passes.

tkelman added a commit that referenced this pull request Mar 15, 2016
[release-0.4] WIP: backports for 0.4.4
@tkelman tkelman merged commit 793526a into release-0.4 Mar 15, 2016
@tkelman tkelman deleted the tk/backports-0.4.4 branch March 15, 2016 12:17
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.