art suite release notes 2.06.00 (2017/02/07)¶
External package changes¶
|Package||art 2.05.00 (previous)||art 2.06.00 (this version)|
Platform/compiler support changes¶
N.B. Due to a late-breaking issue, This release does not support
e10. Support for
e10 will be restored for subsequent releases in the 2.06.X series, but is expected to be removed permanently with the next feature release.
- New qualifier e14: GCC 6.3.0 with "new", GCC5 ABI.
- Qualifer e10 is deprecated: expect support to be removed with the next minor release of the suite.
- Ubuntu 16.04 LTS is now supported (
- Ubuntu 14.04 LTS is deprecated: builds may be produced on request. Expect support to be removed with the next minor release of the suite.
- MacOS Sierra (Darwin 16) is now supported (
e14only) provided SIP is disabled.
- MacOS El Capitan (Darwin 15) with SIP disabled continues to be supported, as does Mac OS X Yosemite (Dawrin 14).
- SLF6 and SL7 (X86-64 architecture) continue to be supported.
- cetlib - Bug #12938: OpenSSL is now no longer directly linked against for macOS builds.
- Some issues related to compiling with Clang, as reported by Ben Morgan, were addressed:
- canvas - Bug #14378: Can not read association in old file.
This problem related not to the reading of an
art::Assns, but rather to the event analysis facilities to obtain the size of such collections.
- art - Bug #14389: --tmpdir and eventdump.fcl don't get along
Specification of the
--tmpdircommand line option now only adds a configuration entry for those modules and services that advertise support for same.
- art - Bug #14430: Art v2_05_00 failed assertion.
assert()that is the subject of this report is erroneous in the specific case, and overly paranoid in the general. It has been removed.
- art - Bug #14664: Ambiguous error type when getProductID fails.
InsertFailureerror category has been split into a
ProductRegistrationFailurecode and a
- art - Bug #14763:
FindManysilently ignores associations to non-available products
- art - Bug #14794: Event viewer crashing with "NULL run" error
- art - Bug #15456:
begin(Sub)Run()called after processEvent in some cases.
This was a long-standing bug in the Boost state chart state machine, which was fixed by the replacement of same.
There is no
e10 build on any platform, due to the inadvertant inclusion of some code which is not supported in GCC 4.9.3.