hal: Branch 'master' - 2 commits
David Zeuthen
david at kemper.freedesktop.org
Sun Apr 1 10:27:52 PDT 2007
NEWS | 2
doc/TODO | 138 +++++++++++++++++++++++++++------------------------------------
2 files changed, 61 insertions(+), 79 deletions(-)
New commits:
diff-tree 6992f57a90e9e2a5b9f9c008204f2ef953d4db63 (from ac8bf1edd90549875c2c0a9d8601a6f8a9d3e8c7)
Author: David Zeuthen <davidz at redhat.com>
Date: Sun Apr 1 13:27:45 2007 -0400
update TODO list
diff --git a/doc/TODO b/doc/TODO
index 844bd19..a429ae0 100644
--- a/doc/TODO
+++ b/doc/TODO
@@ -1,5 +1,4 @@
-
Ongoing items that always need work
-----------------------------------
@@ -12,22 +11,44 @@ Ongoing items that always need work
- Audit all code and drop privileges in helpers where appropriate
-Items specifically planned for 0.5.9
+Items specifically planned for 0.5.10
+(TODO: need to go through and nominate things for 0.5.10)
------------------------------------
+ - Rip out translation framework (.po files) since only deprecated
+ functions in libhal-storage have relied on it since the 0.5.8
+ release.
+
+ - Rip out hal-device-manager for good now that there is a
+ gnome-device-manager on the horizon
+
+ - Release engineering. Would be nice to have a set of Makefile
+ targets to automatically make a new release, upload it somewhere
+ and send announcement mail. Seems that cairo have something we
+ might be able to lift.
+
- Make sure we check if the optical drives supports the commands
we are sending to it (proposed by pjones)
- - Move /usr/libexec to /usr/lib/hal/ - drop /usr/lib/hal/scripts -
- use /usr/lib/hal/methods
+ - Singleton addons (s-addon); the idea that an add-on can attach to
+ several device objects at the same time.
+ - A s-addon would just implement, say, the interface
+ org.freedesktop.Hal.SingletonAddon with methods DeviceAdded(),
+ DeviceRemoved() through the hald direct connection.
+ - An s-addon would be identified by it's commandline; e.g.
+ info.s-addons = {"my-s-addon --foo", "my-s-addon --bar"} would
+ be two distinct addons.
+ - Bluez devs need this to start the Bluez daemon via HAL
+ - Can consolidate other things like hald-addon-keyboard
- - Think hard about the interfaces HAL exports to addons as out-of-tree
- code like NUT now starts to use it. Includes interface stability thoughts
- and what we guarantee and what we don't. Might have license implications
- too.
+ - Move /usr/libexec to /usr/lib/hal/ - drop /usr/lib/hal/scripts -
+ use /usr/lib/hal/methods. Need to provide backwards compat
+ for 3rd party so we need to retain /usr/lib/hal/scripts
- - Finalize the work on org.freedesktop.Hal.Device.VideoAdapterPM
- including the XML file format matching most adapters
+ - Think hard about the interfaces HAL exports to addons as
+ out-of-tree code like NUT and Bluez now starts to use it. Includes
+ interface stability thoughts and what we guarantee and what we
+ don't. Might have license implications too.
- LVM support
https://bugs.freedesktop.org/show_bug.cgi?id=6427
@@ -47,58 +68,54 @@ Items specifically planned for 0.5.9
several hundreds of megabyte so the drive only needs to spin up every
e.g. 20 minutes while watching a DVD movie. This enables the kernel
driver to put the drive in a low-power mode to get substantially better
- battery life.
+ battery life. Probably requires kernel work since the pages would
+ have to be pinned?
- - Provide mechanism for making an application inhibit polling of a drive.
- This is useful combined with the TODO right above. It will probably mean
- that the eject button on the drive is rendered useless but that is
- justified by the DVD player application displays a soft button for ejecting
- the media.
-
- - Use udev data instead of probing (whenever possible)
-
- - Performance! Profile and optimize! Too big of a task to describe in
- one bullet. Feel free to provide patches to this file to break it up
- after some analysis.
+ - Use udev data instead of probing (whenever possible). Here we need
+ the raw serial from udev; for e.g. USB drives, udev prepends it
+ with the make/model of the usb device.
- Configuration of serial and other legacy ports, see
http://lists.freedesktop.org/archives/hal/2006-August/005784.html
- - Devices with device files should export an interface with for
- requesting / dropping access to the device, e.g.
+ - Clean up what the user did by invoking methods on HAL
+ - e.g. unmount storage devices
+ - other things
+ - we can now sanely do this thanks to ConsoleKit
+ - depends on PolicyKit
+
+ - Make HAL unconditionally depend on pm-utils and rip out support
+ for distro-specific tools
+
+ - Devices of capability access_control could export an interface with
+ for requesting / dropping access to the device, e.g.
RequestAccess()
GiveupAccess() (better naming welcome)
which will change device file ownership or add/remove an ACL.
-
-Other items
------------
-
- - It appears Nero is adding empty dummy sessions when leaving the disc
- open, see https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=206669
- Perhaps it would be useful to probe earlier sessions for file systems
- too. Need to look into the various standards a bit more.
+ Depends on PolicyKit
- It would be nice if method calls could return string, bool int etc.
Perhaps use dbus-send style, e.g. int32:<number>' or 'string:the_string'
or 'bool:true'. Granted, this is already possible today through addons
so maybe not important.
- - Finish up and merge mjg59's Bluetooth code
+Other items
+-----------
- Device firmware. It would be nice to export a property and some
DeviceCondition when a device asked for firmware that was missing
and is such in a state where it cannot operate. Might need kernel
- work.
-
- - Take a few days out of the calender and give some love to the spec
+ work. Same goes for drivers.
- Export ink levels for printers
https://bugs.freedesktop.org/show_bug.cgi?id=2295
- Signal when a USB device doesn't have enough power
https://bugs.freedesktop.org/show_bug.cgi?id=4452
+ (it appears that this is true IFF the kernel leaves
+ bConfigurationValue as zero)
- Teach HAL about multisession and partitioned optical discs - may
require reworking the 'fakevolume' code as we'd be adding more
@@ -107,24 +124,6 @@ Other items
Optimization
------------
- - Don't load, parse all fdi files 4 times for every device object
- added. It's wasteful to parse the same file over and over agin
- - load, parse fdi file into "rule object", see
- http://lists.freedesktop.org/archives/hal/2006-July/005672.html
- - load all rule objects at startup into cache
- - evict all cache items after startup
- - expire cache items after timeout / LRU scheme
-
- - Don't load pci.ids, usb.ids into memory (pci.ids 426333 bytes, usb.ids
- is 153540 bytes) at startup (even without parsing it); it costs
- 580K right now. It's wasteful to parse the same file over and over again.
- - load, parse file into efficient lookup tree
- - free tree when no longer used?
-
- - Device Store lookup cleanups
-
- - Device Object property cleanups
-
- Port hal daemon core to use dbus glib bindings
- One proxy per hal device object
- hal device object is an GObject
@@ -136,33 +135,16 @@ Optimization
- Linux: Optionally drop sysfs coldplug code
- for starting up before udev and piggy tailing on udev's coldplug code
+ - Performance! Profile and optimize! Too big of a task to describe in
+ one bullet. Feel free to provide patches to this file to break it up
+ after some analysis.
+
External projects:
------------------
- - We need to finish PolicyKit, at least get 0.2 out and bring it to
- 1.0 rather fast (ideally before Spring 2007).
-
- - D-BUS needs to go to 1.0
-
- - We need secure ways to track user sessions and from what user
- session a D-BUS call into HAL originates from. We also want
- to know when a user session is inactive (e.g. fast user
- switching). This is Very Important, without this we cannot
-
- - clean up what the user did by invoking methods on HAL
- - e.g. unmount storage devices
- - adjust ACL's cf. RequestAccess() and GiveupAccess() above
- - other things
-
- - refuse service to inactive sessions
- - e.g. refuse Suspend() or Hibernate() from the inactive
- user when another user is active on the same console from
- e.g. fast-user-switching
-
- - pm-utils guys need to release their stuff - when a working
- release is out give a grace period of a number of months for
- distros to switch to pm-utils before ripping out the callouts
- to distro specific scripts.
+ - We need to finish PolicyKit, at least get 0.3 out (in Spring 2007)
+ and work on bringing it to 1.0 rather fast (ideally before Fall
+ 2007).
- Runtime power management in the kernel. This is still ongoing
and AFAIK there is no concrete idea about the user space interface.
diff-tree ac8bf1edd90549875c2c0a9d8601a6f8a9d3e8c7 (from e0d3327b68fc043e6ac8a1b0bdf97249476a3968)
Author: David Zeuthen <davidz at redhat.com>
Date: Sun Apr 1 13:27:08 2007 -0400
update hal-info requirement to 20070328
diff --git a/NEWS b/NEWS
index a52bd23..a97fc14 100644
--- a/NEWS
+++ b/NEWS
@@ -11,7 +11,7 @@ Requirements for HAL 0.5.9 ""
- dbus >= 0.61 (with glib bindings)
- glib >= 2.6.0
- expat >= 1.95.8
- - hal-info >= git version as of 20070323
+ - hal-info >= git version as of 20070328
- libusb >= 0 0.1.10a (optional)
- pciutils >= 2.2.3 (optional
- dmidecode >= 2.7 (optional)
More information about the hal-commit
mailing list