Tag a new release?

[jrevels]

Being the swashbuckling Julian that I am, I decided to throw caution to the wind and implement JuliaDiff/ForwardDiff.jl#215 against StaticArrays master rather than latest tag (mainly because I wanted #127).

Would it be possible to tag a new release here, so that I can get Travis passing for JuliaDiff/ForwardDiff.jl#215?

Thanks!

[c42f]

Nice! I think master is in reasonable shape for a tag. Looks like it will be 0.5.0, with #127 a breaking change.

Release notes roughly as follows -

Bug fixes:
#124 (indexing with zero sized arrays)
#139 (Type inference workaround for index_sizes)

Minor feature additions and fixes
#128 (Fixes for additional fancy linear indexing)
#131, #140 (Return SVector from some mixed static/abstract array operations)

Breaking changes:
#127 (Unify SArray, SMatrix, SVector - Tuple must now be passed as an SArray size parameter)

It looks like bd3236a could do with a little more test coverage.

@andyferris do you think we're in reasonable shape to tag?

[c42f]

@andyferris what do you want to do with #134 vs just tagging right away (probably with #138 merged too though).

If we go right ahead with #134, we haven't given any time for stabilization before releasing a big breaking change... which doesn't seem ideal. On the plus side, it means users can get the breakage of #127 and #134 over together, and they are rather related.

[tkoolen]

Just my two cents: I would prefer a 0.4.1 without #134, as this would be the quickest way to get a version of RigidBodyDynamics out that supports Julia 0.6-. I need compatible versions of StaticArrays and Rotations that work on 0.6-, and I need #139.

[andyferris]

My preference is to merge #134 first (or at least get feedback if it is desirable). Otherwise we'd need 0.5.0 and 0.6.0 and yeah, it's related to Keno's work.

[c42f]

Ok I reviewed #134. It seems fairly harmless overall and worth getting merged to avoid saddling users with yet another breaking version.

[c42f]

We need to get this done, I don't want to be blocking people - I'll tag and release tomorrow unless there's any objections. Updated release notes:

---

Release 0.5.0

Breaking changes:

• Unify types so that SVector and SMatrix are aliases for 1D and 2D
  SArray. To make this work, the SArray size parameter must now be passed
  as a Tuple type rather than tuple instance (RFC: Unify SArray,SMatrix,SVector #127).

• Add size information to the base StaticArray type and type aliases, so
  that, for example, StaticMatrix{2,2,T} can be used for directly
  dispatching on 2x2 matrices, without messing with the Size trait (WIP Add size to abstract StaticArray type #134)

Feature additions and fixes:

• Fixes for additional fancy linear indexing (Bug in getindex with colon #128)
• Return SVector from some mixed static/abstract array operations (SVector + Vector: Julia 0.5 vs 0.6 #131, Reinstate StaticMatrix * AbstractVector -> StaticVector #140)
• Bring back StaticArrays.FixedSizeArrays for julia-0.6 (bring back fixedsizearrays #138)

Bug fixes:

• Indexing with zero sized arrays (Fix indexing with zero-size arrays #124)
• Type inference workaround for index_sizes (Workaround for type inference problem in index_sizes #139)
• Fix concatenation of arrays of FieldVector (Concatenating arrays of FieldVectors fails due to ambiguity #146)
• Fix for fancy indexing with a single SMatrix (Restore fancy APL indexing #148)

[andyferris]

Sounds great to me!

[andyferris]

I guess there are a couple of bug fixes remaining?

[andyferris]

@c42f please consider #147 and #148 before tagging - I think they are ready to go.

[andyferris]

There's still #125 but I'm happy to tag first and bugfix that one afterwards.

[c42f]

Nice, thanks Andy, I'll review those asap.

Definitely let's put further bug fixing into the next release... as long as there's not more than the previous release I'm happy ;-)

[c42f]

JuliaLang/METADATA.jl#8999

[jrevels]

Thanks again to both of you for your hard work making such a useful package!