Relicense all of Ingo's BSD/MIT+advertising clause'd code as plain MIT.

Via email:
> I'm fine with re-licensing all code using BSD license + advertising
> clause to MIT license.

Fix catching polymorphic type by value

Replace catching polymorphic type std::bad_alloc 'by value'
with 'by reference'.
Pointed by gcc8

Includes cleanup.


git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@31802 a95241bf-73f2-0310-859d-f6bbb57e9c96

* Fixed GCC 4 build and warnings.
* Used std::vector in BlockingQueue instead of Ingo's Vector until I can figure out the GCC4 problems.
This is untested, but I will test this on another machine in a minute.


git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30423 a95241bf-73f2-0310-859d-f6bbb57e9c96

Sorry, this is a bit of a mess yet and needs more refactoring...
* I've implemented my own copy engine which runs a separate writer thread.
Especially when installing from one hard drive to another, it should speed
things up. When installing from CD, the entire harddisk write time should
at least disappear in the CD read time. When installing from/to the same
drive, the new routine is as fast as the Tracker code that was used before.
* When stopping/canceling the copy process, the copy engine will now actually
block, instead of happily going on while the alert is showing. I've
still implemented that to not leave incomplete files around, like before.
* There is now a progress bar, based on data bytes in files to copy. Reason
being that it's quite fast to collect this info versus also scanning all
attributes.
* The copy engine itself was used in another project before and was tested
with SHA256 hashes of the original files and the copied files, so I think
it can be trusted.
* There is now a lot of code that can be removed or should be cleaned up.
What was the "CopyEngine" should become the "WorkerThread" or something.
* I still had a mental TODO about skipping the swap file (there is none on
CDs, but when copying a harddisk installation there may be one), I just
remembered this when typing this commit message... will fix that next.


git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30394 a95241bf-73f2-0310-859d-f6bbb57e9c96

Includes cleanup.


git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@31802 a95241bf-73f2-0310-859d-f6bbb57e9c96

* Fixed GCC 4 build and warnings.
* Used std::vector in BlockingQueue instead of Ingo's Vector until I can figure out the GCC4 problems.
This is untested, but I will test this on another machine in a minute.


git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30423 a95241bf-73f2-0310-859d-f6bbb57e9c96

Sorry, this is a bit of a mess yet and needs more refactoring...
* I've implemented my own copy engine which runs a separate writer thread.
Especially when installing from one hard drive to another, it should speed
things up. When installing from CD, the entire harddisk write time should
at least disappear in the CD read time. When installing from/to the same
drive, the new routine is as fast as the Tracker code that was used before.
* When stopping/canceling the copy process, the copy engine will now actually
block, instead of happily going on while the alert is showing. I've
still implemented that to not leave incomplete files around, like before.
* There is now a progress bar, based on data bytes in files to copy. Reason
being that it's quite fast to collect this info versus also scanning all
attributes.
* The copy engine itself was used in another project before and was tested
with SHA256 hashes of the original files and the copied files, so I think
it can be trusted.
* There is now a lot of code that can be removed or should be cleaned up.
What was the "CopyEngine" should become the "WorkerThread" or something.
* I still had a mental TODO about skipping the swap file (there is none on
CDs, but when copying a harddisk installation there may be one), I just
remembered this when typing this commit message... will fix that next.


git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@30394 a95241bf-73f2-0310-859d-f6bbb57e9c96