Re: It's Huntin' Season
>>"Nicolas" == Nicolas Boullis <Boullis.Nicolas@libertysurf.fr> writes:
Nicolas> On Fri, Feb 01, 2002 at 04:53:24PM -0600, Manoj Srivastava wrote:
>> If you look at the script I posted, the user is not asked any
>> questions if the file on the system is the same file that is
>> currently being shipped. In the general case, I think this is good
>> enough; if one cares, one can special case the md5sum of the last
>> non-conffile one shipped and check that, but I hate special cases.
Nicolas> I agree that special cases should generally be avoided. But in such a
Nicolas> situation, these special cases help our users to upgrade smoothly.
Nicolas> Hence, I think that, in such a situation, special cases are a
Nicolas> Good Thing (tm).
FUD. Asking a question to recover from a broken package does
not mean an unsmooth upgrade. Indeed, merely treating the transition
from a package theat over writes user changes to one that preserves
them, while noticing that the users current file is different from
the one shipped, and asking them what to do about it, is a _smoooth_
All your check eliminate is a question before replacing the
config file. Allowing people to specify multiple files that may
contain old md5sums is feasible (use a dir instead of a file, and
check every file in that directory), but the added complexity (and
thus potential for error) is not worth the reward (which is a
avoiding a question during postinst)
Offer me reasons why I may be in error, and I'll implement any
number of historical md5sums that'l be checked in addition to the
dynamic md5sum of the ``last'' released version of the config file.
He who has imagination without learning has wings but no feet.
Manoj Srivastava <email@example.com> <http://www.debian.org/%7Esrivasta/>
1024R/C7261095 print CB D9 F4 12 68 07 E4 05 CC 2D 27 12 1D F5 E8 6E
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C