UltraVNC 1.2.0.7 test - Download links

[RudiDeVos]

Test 1
This wil be part of the upcoming 1207 in afew months, to discover bugs early some changes
are now released as test1
download
http://www.uvnc.eu/download/1207/ultrav ... _test1.zip

changes
w8hook fixes

[/http://www.uvnc.eu/download/1206/winvnc_w8test.zip

I added some check to the w8hook to verify if it was proper started.
both winvnc and w8hook need to be replaced to test

added new encoder provided by Adam walling

liblzma/xz cuts bandwidth by about half in my experience, at the
expense of more cpu and memory. which is why i've been hesitant about
pushing it into ultravnc.

but I use it for a singleclick/helpdesk distribution where everything
runs through a repeater and the bandwidth savings are very important
in that case. In some situations the bandwidth usage drops
dramatically since xz can have a very large dictionary (compared to
zlib's 32k limit)

additionally it uses a EncodeBulkRects function, like EncodeRect but
handles multple rects at once -- this way the compression does not
need to flush after every rect.

Please test

[Skyfighter]

tested environment:
- server @ Win8.1 x64 Pro (ultravnc_1207_test1.zip\x64\winvnc.exe + w8hook64.dll on top of installed v1.2.0.6)
- clientA @ Win8.1 x64 Pro (ultravnc_1207_test1.zip\x64\vncviewer.exe)
- clientB @ Win7 x64 Pro (ultravnc_1207_test1.zip\x64\vncviewer.exe)

1) screen updates appear to be laggy / slower (e.g. mouse over effects) compared to v1.2.0.6+winvnc_w8test.zip
2) viewer always crashes on manual selection of encoding (hextile,ultra,u2); it doesn't matter if the manual encoding selection takes place before connecting or while viewer is connected; the only way for a successful connection is using "auto"-mode; (viewer-crash doesn't seem to harm the server-process)

[Skyfighter]

small addition:

3) observed a few times: viewer is successfully connected -> don't do anything with it -> viewer crashes within 5 Minutes (not sure if there is a fixed interval)

[RudiDeVos]

Internal: viewer crash caused by buffer overrun, some pseudoencoding require buffer extention.