I have included the change log from 6.6.34 onwards. 6.6.34 and 35 had a major flaw with cuda handling so don't even think about trying them.

As usual if you find any errors, by all means mention them here, but they really need to be posted on the Boinc_Alpha mailing list.


Change log for 6.6.34

- client: estimated delay was possibly being calculated incorrectly because of roundoff error

- client: add a 1e-6 slop factor in deciding if a resource is fully utilized.

- MGR: finish changes to combined attach wizard

- MGR: cosmetic changes to combined attach wizard, change "Forgot password" link for account managers to a link to AM's home page, add static text

- client: include device number in message describing NVIDIA GPU, and call it "NVIDIA GPU" rather than "CUDA device". fixes #879

- client: tweak messages

- manager: compile fix

- client: when parsing app_info.xml, make sure coprocessors are present. fixes #911

- client: Fix missing argument in printf statements

- GR Installer: Update make_GridRepublic.sh script for new default screensaver

- MGR: Backport a fix to the ProjectListCtrl in the wizard from later branches for Linux distros. Fixes# 759

- fix for coprocessors

- MGR: Accept Unicode input for usernames and password in the manager now that the manager is Unicode enabled. Input will show up as UTF-8 strings in the CC and project. Fixes #452

- MGR: Fix bug in browser cookie detection code where Firefox v3 cookie detection was returning true even for cookies that don't exist.

- MGR: Add some basic validation for account manager cookie detection.

- client: if scheduler request didn't request work, don't report 0 tasks


Change Log for 6.6.35

It may not be so clear from the change log, but the bug of exiting BOINC Manager that kept the client and science applications running has been fixed in this version. Other fixes include:

- GUI RPC: add active_only option to get_results() RPC

- client: read app_info.xml's AFTER scanning GPUs. Otherwise we'll discard all GPU apps and results.

- MGR: Fix compile errors, remove authenticator validity test from detect_cookie_safari()

- MGR: CBOINCClientManager::ProcessExists() finds the process by name if we don't have a pid (Mac, Linux) or process HANDLE (Windows

- Update Mac Manager build instructions for wxWidgets 2.8.10

- scheduler: improve example app plan function for multithread

- MGR: Use code like get_client_mutex() to determine if client is running; KillClient() uses process name if we don't have pid.

- MGR: Implement CBOINCClientManager::KillClient() for Windows.

- MGR: Implement CBOINCClientManager::IsBOINCRunning()for Windows.

- lib: Fix FILE_LOCK::unlock() to reset fd to -1 so FILE_LOCK works correctly with repeated calls.

- MGR: fix CBOINCClientManager::IsBOINCRunning(), add CBOINCClientManager::KillClient(); lib: Fix FILE_LOCK::unlock() for repeated calls

- lib: Fix compiler error on Windows

- MGR: Fix compiler error on Windows

- GUI RPC: include hostid in PROJECT

- Manager: show host ID in project properties

- MGR and Mac MGR: In Async RPC logic, use posix mutexes and condition instead of wxMutex and wxCondition, as this appears to fix an intermittent hang in wxCondition::Wait()

- lib: FILE_LOCK::lock() makes lockfile group-writable so both client and manager can create and write it under sandbox security

- DIAG: Policy change: Only include the BOINC symbol store for BOINC binaries. The old policy was kept around as a backwards compatibility thing before a new flag was added to the diagnostics to identify project applications.


Change log for 6.6.36

For Windows:

- client: read app_info.xml's AFTER scanning GPUs. Otherwise we'll discard all GPU apps and results.
____________
BOINC blog

And just to start the ball rolling...

It seems that 6.6.36 has issues with check pointing. If you stop the client (yes the science apps stop now) and then restart it, the tasks start from the beginning. This is under WinXP.
____________
BOINC blog

It seems that 6.6.36 has issues with check pointing. If you stop the client (yes the science apps stop now) and then restart it, the tasks start from the beginning. This is under WinXP.

Its now the release version for Windows. The Linux version (and Mac OSX?) have gone back to previous "release" version as being recommended.

Just as well too. I wouldn't dream of suspending cuda tasks with anything earlier than 6.6.31 (for Ignasi's thesis work units) as they'd fail.
____________
BOINC blog

I wouldn't dream of suspending cuda tasks with anything earlier than 6.6.31 (for Ignasi's thesis work units) as they'd fail.