Re: lib/54053 (humanize_number(HN_AUTOSCALE) with big buffer doesn't work.)

[Robert Elz <kre%munnari.OZ.AU@localhost>]

The following reply was made to PR lib/54053; it has been noted by GNATS.

From: Robert Elz <kre%munnari.OZ.AU@localhost>
To: Masanobu SAITOH <msaitoh%execsw.org@localhost>
Cc: gnats-bugs%NetBSD.org@localhost, netbsd-bugs%netbsd.org@localhost
Subject: Re: lib/54053 (humanize_number(HN_AUTOSCALE) with big buffer doesn't work.)
Date: Wed, 13 Mar 2019 12:34:50 +0700

     Date:        Wed, 13 Mar 2019 14:10:15 +0900
     From:        Masanobu SAITOH <msaitoh%execsw.org@localhost>
     Message-ID:  <11147520-0be5-fb8a-aa5f-aa4aff29dca9%execsw.org@localhost>
 
   | Yes. It's fixed!
 
 Thanks for confirming.
 
   |   This bug is not serious and it's not required for me to pullup to -8.
 
 OK, in that case I think I won't bother as ...
 
   | But, another person might get the same problem and might waste the time.
   | It would be good to pullup to avoid it and reduce diff between
   | -current and netbsd-[78].
 
 First I think it better to reduce the diff between 8.0 and 8_STABLE
 (and 8.1 when it happens) than between 8 and CURRENT (or 9).  The latter
 is pretty much a waste of time (there's LOTS of diff).
 
 Second, I think it's unlikely that anyone else will ever see this
 problem (and I wonder at just what you were really doing - other than
 the illustration demo included in the PR - when you experienced this
 problem.)   HN_AUTOSCALE and a big buffer together do not really make
 sense - HN_AUTOSCALE is intended to make the number fit in a limited
 field width, if all possible - if the buffer is any larger than (about)
 20 + the length of the suffix to be added, you might just as well simply
 do snprintf() and be done with it, as that is all that humanize_number()
 does in that case.
 
 That will be why, that even though this bug had been there essentially
 forever (decades...) no-one had previously noticed.
 
 I wouldn't bother with a pullup to -7 in any case.  I doubt at this stage
 that there will ever be a new point release for that, and this issue just
 isn't important enough.   It would only get to those running 7_STABLE
 (and upgrading that from time to time) which are a set of people who at
 this stage are probably much better off running 8_STABLE instead.
 
 But if you really want to request a pullup to -8 by all means go ahead.
 (I can supply the needed source-changes message if you don't still have it).
 The patch should apply easily enogh I'd think (and if there were any other
 changes, simply including all of them - pulling -8 up to the same as HEAD
 would not be unreasonable.)
 
 For now, I will just close this though.
 
 kre