Re: Bug#679869: [RFR] templates://fs2ram/{templates}

[Justin B Rye <jbr@edlug.org.uk>]

Philippe Le Brouster wrote:
> Justin B Rye a écrit:
>> Christian PERRIER wrote:
>>> I also wonder whether there should be more explanations about the
>>> benefit one can get from such switch.
>> 
>> I also wonder whether the defaults might flip-flop; maybe it would be
>> safer to point at something like http://wiki.debian.org/RAMTMP... if
>> any such page existed.
> 
> The default value of RAMTMP is probably not the one expected by an user
> of fs2ram (in most use cases, /tmp need to be mounted in ram). Since the
> modification of RAMTMP can't be done automatically, does this justify to
> add a 'note' template in fs2ram ?

I was thinking you could detect whether the relevant directories
currently are mounted as tmpfs and escalate it to an "error" template
if not; but on second thoughts that probably wouldn't be acceptable.

If all the users who like the package description and install the
package are then going to have to be told to follow the setup
instructions in the README, maybe instead of interrupting the install
you should give the warning in the package description:

   This package provides two pre-unmount scripts designed to preserve folder
   structure and file permissions across reboots: this is needed to allow
 - hierarchies suc as /var/cache or /var/log to be mounted as tmpfs.
 -
 + hierarchies such as /var/cache or /var/log to be mounted as tmpfs.
 + .
 + See the README file for instructions on enabling tmpfs mountpoints.

(If there was a README file...)
-- 
JBR	with qualifications in linguistics, experience as a Debian
	sysadmin, and probably no clue about this particular package