[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-devel
Subject:    Re: fsync() madness
From:       Sami Liedes <sliedes () cc ! hut ! fi>
Date:       2008-04-20 18:44:03
Message-ID: 20080420184403.GE8760 () nun4 ! kyla ! fi
[Download RAW message or body]

[Attachment #2 (multipart/signed)]


On Sun, Apr 20, 2008 at 07:08:13PM +0200, Bart Van Assche wrote:
> According to the above thread Konqueror is using KSaveFile for saving
> the browser history, which triggers fdatasync()'s, and this is
> unfortunate. At least with ext3, appending data to a file is
> guaranteed not to corrupt the contents of that file in case of a power
> failure. The new contents is either appended or lost. Would it be a
> good idea to modify Konqueror such that appending a new URL to the
> browser history only involves appending data to a file, and previously
> written data is not modified ? The interval at which data is written

Or then just do it whatever way and in the very unlikely event of very
unfortunately timed power loss (you should be using UPS anyway) suffer
the losses like pretty much every other piece of software written for
Unix/Linux.

This very case is in my opinion quite illustrative. We're talking
about _browser_history_, not some nuclear attack plans. It's hard to
see how it can be that important, while all other normal software just
trusts the kernel and the fs and performs fine. At least behaving like
good Unix citizens are expected to should be made an option, and I
think it should be on by default.

	Sami

["signature.asc" (application/pgp-signature)]

>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<


[prev in list] [next in list] [prev in thread] [next in thread]