Hi Michael, and thanks for your care about usb-modeswitch-data,

On Thursday 17 March 2011 07:29:48 Michael Biebl wrote:
> With today's upload the files were moved out of /etc/ into /usr/share.
> Have you read Josua's comment, that those files are needed during boot?

Yes. I am in regular contact with Josua, both before and after upstream 
releases and we keep us cross-informed of potential issues arising on both
sides.

> With the files in /usr/share, you will run into problems if /usr is on a
> separate partition because udev is run rather early before boot (before
> other partitions are mounted).
> In case your modem is already attached during boot, it won't be switched,
> unless you plug it out and in again.

Josua covered this already, so I won't repeat his words here. From my point of
view: "It Just Works"™.

Are you experiencing "devices plugged at boot time not switched when the boot
ended" ? If you are, it's a bug that you should really report (against usb-
modeswitch). (I don't think you are… :-) ).

> Have you considered that?

Yes. I think both Josua an I did consider that, carefully. And the result is
working.

In another mail, on Thursday 17 March 2011 08:31:20 you wrote:
> Looking at /lib/udev/usb_modeswitch though, it seems you used wait_for_file
> from /lib/udev/hotplug.functions. TTBOMK, this is a Debian/Ubuntu specific
> feature. I quickly checked a Fedora F14 and openSUSE 11.4 installation.
> Neither of those has has a hotplug.functions file and wait_for_file
> function.

Exactly. And that's because I an patching Josua's /lib/udev/usb_modeswitch :
http://patch-tracker.debian.org/patch/series/view/usb-modeswitch/1.1.7-1/03_use_udev_specifics.patch

So the usage of wait_for_file _is_ Debian-specific (Ubuntu added the
/lib/udev/hotplug.functions partly because of my patch) and limits the code
duplication around udev, while keeping the "timeout" definition in one place
only.

Is all this clear enough now or do you need more explanations ? I'd be glad to
guide you trough more aspects of the packaging (and I'm sure Josua would do
the same for the upstream code).

Best regards, 

OdyX

Attachment: signature.asc
Description: This is a digitally signed message part.

Reply via email to