Mir

Merge lp://staging/~robertcarr/mir/disable-sequences-and-add-terminate-handler into lp://staging/~mir-team/mir/trunk

disable-sequences-and-add-terminate-handler
Merge into trunk

Proposed by Robert Carr on 2013-05-15

Status:	Rejected
Rejected by:	Robert Carr on 2013-05-28
Proposed branch:	lp://staging/~robertcarr/mir/disable-sequences-and-add-terminate-handler
Merge into:	lp://staging/~mir-team/mir/trunk
Diff against target:	278 lines (+118/-3) 9 files modified examples/demo-shell/demo_shell.cpp (+33/-2) include/test/mir_test_doubles/null_virtual_terminal.h (+1/-0) src/server/graphics/gbm/gbm_display.cpp (+1/-0) src/server/graphics/gbm/gbm_platform.cpp (+14/-0) src/server/graphics/gbm/linux_virtual_terminal.cpp (+11/-0) src/server/graphics/gbm/linux_virtual_terminal.h (+2/-0) src/server/graphics/gbm/virtual_terminal.h (+1/-0) tests/unit-tests/graphics/gbm/test_gbm_display.cpp (+4/-1) tests/unit-tests/graphics/gbm/test_linux_virtual_terminal.cpp (+51/-0)
To merge this branch:	bzr merge lp://staging/~robertcarr/mir/disable-sequences-and-add-terminate-handler
Related bugs:	Link a bug report

Reviewer	Review Type	Date Requested	Status
Alan Griffiths			Needs Fixing on 2013-05-20
Alexandros Frantzis (community)		2013-05-15	Needs Fixing on 2013-05-16
PS Jenkins bot (community)	continuous-integration		Approve on 2013-05-15
Review via email: mp+164014@code.staging.launchpad.net

Commit message

Disable control sequences (ctrl+c, etc...) and add a way to kill mir (demo shell) in lieu of this.

Description of the change

Disable control sequences (ctrl+c, etc...) and add a way to kill mir (demo shell) in lieu of this.

Revision history for this message

PS Jenkins bot (ps-jenkins) wrote on 2013-05-15:

PASSED: Continuous integration, rev:692
http://jenkins.qa.ubuntu.com/job/mir-ci/589/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-raring-i386-build/696
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-clang-raring-amd64-build/579
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-raring-amd64-ci/73
        deb: http://jenkins.qa.ubuntu.com/job/mir-raring-amd64-ci/73/artifact/work/output/*zip*/output.zip
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-vm-ci-build/./distribution=precise,flavor=amd64/318

Click here to trigger a rebuild:
http://s-jenkins:8080/job/mir-ci/589/rebuild

review: Approve (continuous-integration)

Revision history for this message

Robert Ancell (robert-ancell) wrote on 2013-05-16:

The name disable_control_sequences() seems confusing - this is just setting the console into raw mode right? Shouldn't it be called set_raw_mode() or similar?

Is there an easy way for the demo shell to note that it can only be quit using alt+ctrl+backspace? This is likely to confuse people using this.

This change is going to affect all shells, not just the demo shell. It might be worth putting alt+ctrl+backspace into libmirserver by default (at least for now) so other shells also are quittable if we disable input.

Finally, check with Alexandros but I think there was some issues that meant he hadn't enabled raw input yet.

Revision history for this message

Alexandros Frantzis (afrantzis) wrote on 2013-05-16:

> This change is going to affect all shells, not just the demo shell.
> It might be worth putting alt+ctrl+backspace into libmirserver by default
> (at least for now) so other shells also are quittable if we disable input.

+1, we can put it in run_mir.cpp

> Finally, check with Alexandros but I think there was some issues that meant
> he hadn't enabled raw input yet.

My only concern is that we would disable VT switching key sequences, while we didn't yet have another way to handle them. It seems that the current raw mode options don't affect this, so no problem there.

Plus, I think that the input stack has matured enough to support this (i.e. catching ctrl-alt-f1 and initiating a VT switch), and when we get this we can completely disable VT input (something like ioctl(fd, KDSKBMODE, K_OFF), but I need to experiment), which means that we won't need FOPS::make_raw()/VT::disable_control_sequences() any more, we will just need VT::disable_input().

137 +void mgg::LinuxVirtualTerminal::disable_control_sequences()

We also need to restore the VT input handling state in the LinuxVirtualTerminal destructor, to get back to a sane console after shutdown.

review: Needs Fixing

Revision history for this message

Alan Griffiths (alan-griffiths) wrote on 2013-05-20:

In addition to Alexandros's comments around VT switching there should be some *documentation* to alert users that:

1. Ctrl-C doesn't kill mir_demo_shell
2. How to kill it

review: Needs Fixing

Unmerged revisions

692. By Robert Carr on 2013-05-15

Correct terminate handler

691. By Robert Carr on 2013-05-15

Make test pass

690. By Robert Carr on 2013-05-15

Failing test for disabling control sequences

689. By Robert Carr on 2013-05-15

Add a disable control sequences interface to VT

688. By Robert Carr on 2013-05-15

Add a make_raw method to VTFileOperations which disables control sequences

687. By Robert Carr on 2013-05-15

Allow access to the internal egl native display without usage of a surface.

Approved by PS Jenkins bot, Kevin DuBois.

686. By Robert Ancell on 2013-05-15

Release version 0.0.3

685. By Alexandros Frantzis on 2013-05-14

shell: Close all open sessions when destroying the SessionManager

We need to do this manually to break the cyclic dependency between msh::Session
and msh::InputTargetListener, since our implementations of these interfaces
keep strong references to each other. This change fixes LP:#1170643. Fixes: https://bugs.launchpad.net/bugs/1170643.

Approved by Robert Carr, Alan Griffiths, PS Jenkins bot.

684. By Alan Griffiths on 2013-05-14

Remove MIR_GCC_VERSION variable.

This was useful when the Android SDK wasn't up to date. We're now building
on a full Raring/Saucy image, so we don't want to use a non-default compiler.

Approved by Kevin DuBois, Alexandros Frantzis, PS Jenkins bot.

683. By Chris Halse Rogers on 2013-05-14