If m_current.block is filled and connection is switched to
another host and protocol, the last r...

Merge from ethminer

Allow setactiveconnection by host pattern match in API

- By default log goes to stderr to maintain backwards compat.

- strictly reformatting, no functional changes

Removed the reciprocal negation.
In addition some slight adjustments to SimulateClient

clang-format cuda code

Optionally send log output to stdout

- By default log goes to stderr to maintain backwards compat.

- strictly reformatting, no functional changes

Some small fixes to AL-Lot-of-stuff

Bump version: 0.18.0-alpha.0 → 0.18.0-alpha.1

`m_securesocket != nullptr` already implies `m_conn->SecLevel() != SecureLevel::NONE`

NFC: Format source using clang-format.

Al lot of stuff

If no devices found ethminer returns -1
Change arguments to --help to make travis happy.

Clean after tentative merge to master

Import work from AL

Remove one multiply.

Bump version: 0.17.0-alpha.1 → 0.18.0-alpha.0

Bump version: 0.17.0-alpha.1 → 0.17.0-rc.0

Revert commit a8bb4f86b1ead859e0b9dc38dc37cd0225651b99

- Fix logic preventing proper parse of URI fragment
- Rename String method to conventional str f...

This reverts commit 019ad03447f59b28f3c5b6568b581d9c62a30d49.

fixed travis build

Minor cleanups in Ethash Cuda

Pci devices abstraction

EthStratumClient: Small cleanups

It's already set by setTarget()

Instead of using indexes make certain every GPU matches it's hw monitor by Pci Id

Allow passing "." as username or workername using %2e.

Fix commit a608a0d481213342db61c2296d8cc146e987ac3e .

As we split username and workername by "." we have to do it before
an urlDecode() is called as o...

Allow passing "." as username or workername using %2e.

As we split username and workername by "." we have to do it before
an urlDecode() is called as o...

Create only one epoch context for all miners instead of having each one to call ethash function.

Enable init() at Miner.cpp level so we can serialize initialization among ALL miners (CUDA and O...

It's only implemented in CUDA and need to reworked for PCI devices abstraction.

It's necessary FormattedMemSize to return the whole string as single element instead of two chun...

Sticking to diff (derived by boundary) expressed in "K megahashes" does not fit when there are v...

Partly amends https://github.com/ethereum-mining/ethminer/issues/1689

Now totally async without the need of an own thread.
Also differentiates among socket errors and...

Hash rate submission is a duty of poolmanager, not a duty for the connector which only have to b...

Measurement of stales at miner's side has no meaning and in 99% of cases does not match with sta...

Treat it as string: no need to convert back and forth to h256

As described in https://github.com/ethereum-mining/ethminer/issues/1681 the computation of targe...