@nakihito T634 landed! Good to go. :-)

(Note this brings us up-to-date with Core on this code, aside from some renames later on)

Failed attribution

To make it easy to compare: https://github.com/bitcoin/bitcoin/compare/34d68bf3a~1..f90c3a62f

Yep, looks like accidental backport in D936.

Seems like quite some follow-ups have occured to this:
https://github.com/bitcoin/bitcoin/pull/10133
https://github.com/bitcoin/bitcoin/pull/11022

For fun, I tried running the script and picked up some changes:

nice!

Notes from core's upgrade: https://bitcoin.org/en/release/v0.17.0

thanks @jasonbcox good points, I've updated description accordingly

BTW how this works: if you specify -blocksdir=/home/mark/spam then:

In D2939#69573, @markblundeberg wrote:

On further thought I think manual testing would be a good idea here, since /in theory/ it's possible that some codebase divergence means there could be a GetDataDir() somewhere else that needs to be changed over.

rebase & did test on testnet

style

In D2994#70355, @jasonbcox wrote:

I can't find the comment right now, but we had a short convo about not adding this at this time. The value of doing so isn't clear beyond estimating IBD time in the GUI. If that's true, we already know it to be quite inaccurate, so doing this just adds work to our release process.

@deadalnix Presumably this is based on PR9497 https://github.com/bitcoin/bitcoin/pull/9497 ?

first constructor to be removed in PR 15108 later on

Looks like replaces_txid is no longer used but replaced_by_txid has one mention in codebase. (wallet.cpp)

note This code will be removed in PR 14955 but that probably won't be backported for a while.
https://github.com/bitcoin/bitcoin/pull/14955

"useless scopes" but in fact they do have a use: allowing easy backports =D

Also includes re-adding a code block because it conflicts with every single backport in this function.

see T634

A note: Bizarrely, in the POSIX form it's not clear whether the string will be nul-terminated if it happens to be exactly buf-sized. https://chromium.googlesource.com/chromium/src/base/+/master/posix/safe_strerror.cc#78 )

Open bug related https://github.com/bitcoin/bitcoin/issues/14466

Abandoning this concept. In fact the BasicTestingSetup constructor sets up several important things.

leave test-subject headers in first header block

fix qt test too

wasn't building qt on my build machine so I missed the qt tests ...

On further thought I think manual testing would be a good idea here, since /in theory/ it's possible that some codebase divergence means there could be a GetDataDir() somewhere else that needs to be changed over.

Note, there are some future backports that should be done relating to this (but can't be done yet), see T636 description.

checked with the respective files & confirm this resolves inconsistency.

Note from @dagurval on the slack--

the fix PR12806 and subsequent qa changes require T568

T636 is needed for the new CheckDiskSpace mechanics to be upgraded cleanly