INCLUDE_CONFIG_FILE in GENERIC
ndenev at gmail.com
Thu Jan 14 07:32:25 UTC 2010
On 14 Jan, 2010, at 24:04 , M. Warner Losh wrote:
> In message: <201001131633.09669.jhb at freebsd.org>
> John Baldwin <jhb at freebsd.org> writes:
> : On Wednesday 13 January 2010 3:36:26 pm Doug Barton wrote:
> : > On 1/13/2010 12:15 PM, John Baldwin wrote:
> : > > On Wednesday 13 January 2010 1:48:38 pm Doug Barton wrote:
> : > >> To address the other responses, Tom, sorry, your suggested text doesn't
> : > >> address my concern. John, I don't think that users would somehow
> : > >> magically know to look in NOTES for more information about an option
> : > >> that is already in GENERIC.
> : > >
> : > > You really think users do not already know to look in manpages or NOTES to
> : > > find out more details about kernel options?
> : >
> : > That's not what I said.
> : <quote>
> : I don't think that users would [..] know to look in NOTES for more information
> : about an option that is [...] in GENERIC.
> : </quote>
> : That seems really straight forward to me, or my English isn't good. I do
> : think users "would know to look in NOTES for more information about an option
> : that is in GENERIC".
> Agreed. That's why I did what I did: I conformed to the usual practice.
> : > > Put
> : > > another way, what makes 'INCLUDE_CONFIG_FILE' sufficiently special that it
> : > > deserves special treatment relative to other kernel options?
> : >
> : > Because the default behavior (not including the actual file) for the
> : > option is contrary to user' reasonable expectation of how the option
> : > should work .... and now I'm repeating myself.
> : I think a better change would be to just change the default behavior of
> : config(8) to do the reasonable thing.
> -C should be the default, and we should invent a new
> '--smaller-saved-config' option.
> : > Seriously, don't you have anything better to do than argue against
> : > including a comment in a config file? I know I do. What is the
> : > overwhelming horror that will arise here if there are more comments
> : > GENERIC than you deem to be absolutely necessary?
> : What is the overwhelming horror about keeping a file readable and allowing
> : users to find extended documentation for INCLUDE_CONFIG_FILE in the same place
> : that they find extended documentation about every other kernel option?
> Yes. That's why I did what I did: to keep things readable.
> : > And yes, I read the part of your message that I snipped about "why do we
> : > have these documents if users don't read them?" The answer is, that's
> : > why I'm suggesting a comment that tells users what man page to read.
> : I think adding comments that merely redirect the users to further
> : documentation only serves to obfuscate. Left unchecked this approach will
> : render files such as GENERIC with a very low signal-to-noise ratio making it
> : harder to parse in a "big picture" way.
> Basically, I'm annoyed too: Our users aren't idiots, and we shouldn't
> be treating them as such at every turn. If there are surprises with
> how INCLUDE_CONFIG_FILE behaves, we should work to make it better, not
> paper over it with a comment.
I just want to add a user's point of view :
To me INCLUDE_CONFIG_FILE sounds like the
whole config file will be included,
not just the output after preprocessing.
So I was thinking about something like two
different options, one "INCLUDE_CONFIG_FILE"
which includes the whole file with comments,
and the other to be just "INCLUDE_CONFIG".
I think these would be pretty self-explanatory.
Yes, it adds another kernel option, but having options
to kernel options looks even more cryptic :)
More information about the svn-src-head