[Docs] First-draft release notes for 0.11.2RC1

[harding]

FIXMEs:

• This mentions that a new libblkmaker is needed but has FIXME for the version number. (@luke-jr knows about this via IRC)
• @petertodd and @maaku should really review the sections, respectively, about CLTV and BIP113 as my descriptions of them are probably flawed.
• Additional contributions and security research doesn't mention anyone specifically; if it should, please tell me who or update yourself at merge time.

Please feel free to make any other changes at merge time.

For reference, I generated the commit-level changelog from here: v0.11.1...v0.11.2rc1

And to make sure I didn't miss anything, I documented everything but the merge commits. Here's the complete changelog in the same order as that compare link above. Commits in bold are the ones I omitted from this pull because they didn't seem to affect behavior. (Commits not in bold are included in these notes.)

• CLTV IsSuperMajority() soft-fork, rebased against v0.11.0 #6707 684636b Make CScriptNum() take nMaxNumSize as an argument
• CLTV IsSuperMajority() soft-fork, rebased against v0.11.0 #6707 6ec08db Move LOCKTIME_THRESHOLD to src/script/script.h
• CLTV IsSuperMajority() soft-fork, rebased against v0.11.0 #6707 4fa7a04 Replace NOP2 with CHECKLOCKTIMEVERIFY (BIP65)
• CLTV IsSuperMajority() soft-fork, rebased against v0.11.0 #6707 6ea5ca4 Enable CHECKLOCKTIMEVERIFY as a standard script verify flag
• CLTV IsSuperMajority() soft-fork, rebased against v0.11.0 #6707 5e82e1c Add CHECKLOCKTIMEVERIFY (BIP65) soft-fork logic
• CLTV IsSuperMajority() soft-fork, rebased against v0.11.0 #6707 c5a27f4 Add RPC tests for the CHECKLOCKTIMEVERIFY (BIP65) soft-fork
• CLTV IsSuperMajority() soft-fork, rebased against v0.11.0 #6707 70a427b CLTV: Add more tests to improve coverage
• CLTV IsSuperMajority() soft-fork, rebased against v0.11.0 #6707 ba1da90 Show softfork status in getblockchaininfo

• 6707 6af25b0 Add BIP65 to getblockchaininfo softforks list

• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 9b9acc2 Fix spelling of Qt
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 01878c9 Fix locking in GetTransaction
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 b3eaa30 [Qt] Raise debug window when requested
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 1e672ae Debian/Ubuntu: Include bitcoin-tx binary
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 2394f4d Debian/Ubuntu: Split bitcoin-tx into its own package
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 6fd0019 Drop "with minimal dependencies" from description
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 a33cd5b [trivial] Fix rpc message "help generate"
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 87a797a build: Remove dependency of bitcoin-cli on secp256k1
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 33d6825 Bugfix: Allow mining on top of old tip blocks for testnet
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 9e45157 build: disable -Wself-assign
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 bfc6154 [Trivial] Fixed typo when referring to a previous section in
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 54f9dee Update bluematt-key, the old one is long-since revoked
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 e42bf16 Clarification of unit test build instructions.
• Update LevelDB tree to include #6917 (0.11) #6945 09a00a1 Add historical release notes for October 2015 bugfix releases

• 6945 21e58b8 build: make sure OpenSSL heeds noexecstack

• Update LevelDB tree to include #6917 (0.11) #6945 0720324 Make fee aware of min relay in pruning.py RPC test
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 4fbfebe Correct spelling mistakes in doc folder
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 7ce2c91 Update debian/changelog and slight tweak to debian/control
• Backport bugfixes to 0.11 (2015-10-22 / f2c869a) #6825 131d7f9 Change URLs to https in debian/control

• 6825 af6edac alias -h for --help

• 6945 95a5039 Set TCP_NODELAY on P2P sockets.

• Update LevelDB tree to include #6917 (0.11) #6945 dfe55bd Do not allow blockfile pruning during reindex.
• Backport #6566, median-past locktime, rebased against 0.11 #6884 a1d3c6f Add rules--presently disabled--for using GetMedianTimePast as end point for lock-time calculations

• 6884 f720c5f Enable policy enforcing GetMedianTimePast as the end point of lock-time constraints

• 6917 0af5b8e leveldb: Win32WritableFile without memory mapping

• 6945 70de437 Update LevelDB

• 6948 4e895b0 Always flush block and undo when switching to new file

[harding]

929b2c7 implements the changes suggested by @gmaxwell (with some minor copyedits) and adds @btcdrak to the credits for his work backporting #6884

I'll be happy to squash once the FIXMEs are addressed (or feel free to squash at merge time).

[sipa]

Perhaps it's worth mentioning that a bug which caused frequent database corruption on Windows was fixed?

[sipa]

Block v4 graphs: http://bitcoin.sipa.be/ver-50k.png and http://bitcoin.sipa.be/ver-2k.png

[gmaxwell]

@sipa Yea, the leveldb fix change is big news for a lot of people. Unfortunately we don't have the AV robustness (chainstate obfsucation) so we can't say that ALL known common sources of corruption on windows are gone.

[harding]

Commit 98d77c5 links to @sipa's graphs and adds a note about the Level DB change to the Notable changes section.

[btcdrak]

Need a release version from @luke-jr for libblkmaker and replace FIXME

[harding]

@btcdrak agreed. This is listed in the FIXME checklist at the top of the PR.

[sipa]

Reduces delay? The change likely makes it slower, but it no longer relies on memory-mapped files for writing, which seems to reduce the problem. We don't know why that is, but it's certainly not impossible that the mmap code we were using was buggy.

[harding]

Ah, that makes more sense. I'll update it. Thanks.

[petertodd]

ACK the BIP65 and BIP113 notes.

IMO you did a better job describing the changes then I did in git master; we should probably consider copying your text when we're closer to releasing v0.12.0

[laanwj]

Thanks for doing this!

It's a bit unfortunate that #6825 (as well as #6707) doesn't refer to the individual original pull numbers of the backported fixes. This confuses my administration - and referring to a pull like "backport fixes" in the release notes isn't that useful. But we can live with that, better next time.

@sipa Yea, the leveldb fix change is big news for a lot of people. Unfortunately we don't have the AV robustness (chainstate obfsucation) so we can't say that ALL known common sources of corruption on windows are gone.

It's not that I didn't try: #6919. But it was to controversial for a backport apparently.

[gmaxwell]

@laanwj Indeed, I was mostly pointing out that there is one known common remaining source of corruption for harding's benefit, and not in any way attempting to impugn the incredible work you've done.

[laanwj]

Went ahead and merged this so that there are some release notes in the repository to refer to in my announcement, feel free to improve this later

[harding]

@laanwj I can dig up the the original information on the backported fixes. Can you clarify how you usually present that information? For example, if the original PR is 1111, the original commit is aaaaaaa, the backport PR is 9999, and the backport commit is fffffff, how would you like that information presented:

• (what we have now) #9999 fffffff Commit title
• (all original info) #1111 aaaaaaa Commit title
• (original PR/backport commit) #1111 fffffff Commit title
• (all the info) #1111 aaaaaaa (#9999 fffffff) Commit title

[btcdrak]

what about the FIXME version numbers for libblkmaker?

[laanwj]

@harding Usually there would be no backport PR, as I cherry-pick them straight into the branch. If there is one, and you have the real PR number available, I'd say link that instead of the backport PR.