Alexander E. Patrakov wrote:
Matthew Burgess wrote:
Alexander E. Patrakov wrote:
Anyway, let me introduce udev-100 first - it (unfortunately) has the
original non-educational CD-ROM rule generator in the recommended set
of rules. The voting suggestion above holds only if we decide to
ignore it, preferring Jim's viewpoint (in his opinion, a program that
writes rules for itself is wrong). I disagree with that opinion, but
don't want to force my disagreement upon others. Additionally, our
work on showing the users the two styles of persistent rules
(identity persistence and location persistence) is going to be lost
if we adopt the stock generators.
Can we not have our cake and eat it too? By which I mean, can we not
use the stock generators explaining which of the styles of the
persistent rules it uses and also showing an example of the other style?
Let's see if I understand you correctly.
1) Explain that "location persistence" and "identity persistence" are
possible
Yes.
2) Show both examples for both CD-ROM and network (i.e., 4 examples
total, as in the current book)
Well, I'd actually be happy with just two examples. i.e. following the
description of what location persistence is, tell folks that an example
is implemented in write_cd_rules (with an explanation of what that
example does, and what the result of the rule will be). Then do the
same with the identity persistence, pointing to write_net_rules.
3) Explain that it is possible to set up scripts that automatically
write persistent rules for a device upon its first discovery, explain
why retrying failed uevents is important for the scripts to work.
Yes.
4) Optionally install pre-written scripts that achieve location
persistence for CD-ROMs and identity persistence for network interfaces.
Yes. When (note, *not* if!) we get the explanatory material correct for
steps 1-3, this should give folks enough information to make a
reasonable decision of whether or not to install the pre-written scripts
or to write their own persistence rules.
How should this be divided into sections?
I don't know what the actual section/subsection numbers would be, but
something like:
1. What is device persistence?
1.1. Identity persistence?
1.1.1. Description of what it is, why/when you would want to use it.
1.1.2. Worked example (taken from write_net_rule)
1.2 Location Persistence
1.2.1. Description of what it is, why/when you would want to use it.
1.2.2. Worked example (taken from write_cd_rule)
1.3. Automatically generating rules - discuss the reasons why you might
or might not want to do this. Provide commands to install the
extras/rule_generator scripts.
So, assuming that we pre-populate the database in the scripted case or
solve the "which card becomes eth0" problem by other means, this covers
the following setups (let's talk about network interfaces for simplicity):
A) Location persistence, unscripted
B) MAC address persistence, unscripted
C) MAC address persistence, scripted
Sounds good enough to me. As far as solving the "which card becomes
eth0" problem, I don't know enough about it at the moment to comment.
I'll try to get udev-100 working here over the weekend. Would you mind
if I tackle this particular upgrade, Alexander (obviously with your
input/advice)? The reason I'd like to do it is to try and improve my
own knowledge of udev.
Thanks,
Matt.
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
