it was already decremented when the post was hidden
- Added a function to forum_repair.php to fix inconsistencies
due to the above bug
svn path=/trunk/boinc/; revision=15433
the ones we are expecting.
- WINSETUP: Do not allow the data directory and install
directory to be the same directory.
- WINSETUP: Reduce the amount of time allocated to
shutdown the BOINC Manager from 30 secs to 5 secs
before killing it. Otherwise the user thinks the
install process is hung.
- WINSETUP: Change text for describing the protected
mode install
- WINSETUP: Add text describing the reason for a reboot
- WINSETUP: Add text describing the need to run a repair
in the case where a third party BOINC CC was installed
or a private drop of the CC was used.
win_build/installerv2/
BOINC.ism
win_build/installerv2/redist/Windows/src/boinccas/
boinccas.rc
boinccas.vcproj
boinccas95.def
boinccas95.vcproj
CAShutdownBOINCManager.cpp
CAValidateInstall.cpp, .h (added)
CAVerifyInstallDirectories.cpp, .h (added)
win_build/installerv2/redist/Windows/Win32/
boinccas.dll
boinccas95.dll
win_build/installerv2/redist/Windows/x64/
boinccas.dll
boinccas95.dll
svn path=/trunk/boinc/; revision=15425
- change ERR_CHILD_FAILED TO EXIT_CHILD_FAILED
(it's an exit code, not a function error code)
- client: reduce severity of benchmark error
- scheduler: compile fix
svn path=/trunk/boinc/; revision=15423
The problem: the variable ordered_scheduled_results
has a list of pointers to RESULTs.
These pointers become dangling when the project is detached.
It's not sufficient to do schedule_cpus() because this returns
immediately if # projects == 0.
So clear the list directly.
Rom: please port to 6.2
svn path=/trunk/boinc/; revision=15407
This supports apps that can do variable amounts of computing;
they can boinc_finish() if their deadline is near.
Rom: please back-port.
svn path=/trunk/boinc/; revision=15395
(fix links on profile-index pages)
- web: use cache-control "no-store" only for home.php.
This makes it so that, e.g. if you scroll to the middle of a long page,
follow a link, then click Back,
you'll go back the original scroll position rather than top of page.
svn path=/trunk/boinc/; revision=15391
There are two aspects:
1) undoing magic quotes (if it's being used).
This must be done for all GET and POST string inputs.
It is now done automatically by get_str() and post_str().
The places that refer to $_GET and $_POST directly
must do it themselves using undo_magic_quotes().
2) Escaping user-supplied strings before using them in DB queries.
This is done by process_user_text()
(which should be renamed db_escape_string()).
The new principle:
call process_user_text() in the function that does the DB query
(not at any higher level).
svn path=/trunk/boinc/; revision=15389
init_data.xml was writing 'symstore' and the diagnostic
code was looking for 'project_symstore'. Why did it take
so long to find? Oh well, make it obvious if it ever
happens again by dumping the symstore url in the crash
report.
Ugh...
lib/
diagnostics.C
diagnostics_win.C
svn path=/trunk/boinc/; revision=15378
1) it's not correct to apply stripslashes() to all form input.
That should be done only if magic quotes is on.
I fixed this by adding a function undo_magic_quotes()
that does the right thing.
2) There's no reason to strip backslashes from user-supplied text
(forum titles and bodies, private messages, passwords etc.)
I'm not sure why this was being done.
It prevented messages from containing backslashes.
svn path=/trunk/boinc/; revision=15364
These are more than 24 lines, so in the VMWare BOINC VM
they scroll off the window and are lost.
Instead, write them to a file (project/project.readme).
Also, only write the message about example app if -test_app
svn path=/trunk/boinc/; revision=15358
David Anderson