Re: bin/50092: huge memory leaks in vi(1) when changing screen size

[Rin Okuyama <okuyama%flex.phys.tohoku.ac.jp@localhost>]

The following reply was made to PR bin/50092; it has been noted by GNATS.

From: Rin Okuyama <okuyama%flex.phys.tohoku.ac.jp@localhost>
To: Julian Coleman <jdc%coris.org.uk@localhost>, gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: bin/50092: huge memory leaks in vi(1) when changing screen size
Date: Tue, 28 Jul 2015 00:10:55 +0900

 Thank you for your comment.
 
 > If we resize many times, we probably only care about the final size.  So,
 > we really could ignore any apart from the last resize.  However, if we're
 > already processing a resize, we should finish processing so that everthing
 > is consistent (i.e. not interrupt the current resize).
 
 Exactly.
 
 > I wonder if we can just save the new size when we're already processing and
 > check at the end of the current processing if a new (and different) size is
 > saved?  Then we would re-run the resize with the new saved size.
 
 I think that your suggestion is a right way to deal with the problem.
 Theoretically, we can do so by switching signal handlers appropriately.
 Let me think for a while about actual implementation.
 
 Except this problem, my patch improves the situation. At least, vi no
 longer wastes MB of memory. Other commands using terminfo/curses, less,
 more, rogue, tetris, and sysinst work fine. Could anyone please review
 and commit it?
 
 Rin