Bug#596038: Writes temporary files all over the filesystem

2010-09-15 Thread martin f krafft
also sprach Nicolas Valcárcel Scerpella [2010.09.16.0720 +0200]: > Can you please test it and give you ok to upload it to the > archive? Not this month, but in October. Sorry. -- .''`. martin f. krafft Related projects: : :' : proud Debian developer http://debian

Bug#596038: Writes temporary files all over the filesystem

2010-09-15 Thread Nicolas Valcárcel Scerpella
Martin, Can you please test it and give you ok to upload it to the archive? -- Nicolas Valcárcel Security Engineer Custom Engineering Solutions Group Canonical OEM Services Mobile: +511 994 293 200 Key fingerprint = 5C4D 0C85 D9C0 98FE 11F9 DD12 524E C3CD EF58 4970 gpg --keyserver keyse

Bug#596038: Writes temporary files all over the filesystem

2010-09-15 Thread David Lutterkort
On Wed, 2010-09-15 at 18:53 +0200, martin f krafft wrote: > also sprach David Lutterkort [2010.09.15.1756 +0200]: > > To catch your use case in all generality, I could just write the > > changed file to a memstream, do the compare from that, and only > > write to .augnew when something has changed

Bug#596038: Writes temporary files all over the filesystem

2010-09-15 Thread martin f krafft
also sprach David Lutterkort [2010.09.15.1756 +0200]: > To catch your use case in all generality, I could just write the > changed file to a memstream, do the compare from that, and only > write to .augnew when something has changed. Except that will break for large files (however unlikely that m

Bug#596038: Writes temporary files all over the filesystem

2010-09-15 Thread David Lutterkort
Martin, you are right - in this case, there's a simple way to avoid writing anything, since Augeas can just check whether the old and the new value for the tree node are different, and hence can figure out if anything needs to be written at all. There are more complicated cases of a noop, like re

Bug#596038: Writes temporary files all over the filesystem

2010-09-14 Thread martin f krafft
also sprach Nicolas Valcárcel Scerpella [2010.09.13.2030 +0200]: > I don't understand the use case of this, why would you want to 'save' > and that the changes aren't written to the disk? whouldn't that be > unusefull? The use case is puppet, which seems to use the equivalent of -n and then comp

Bug#596038: Writes temporary files all over the filesystem

2010-09-14 Thread Patrick
On Sep 14, 2010, at 12:07 AM, Dominique Dumont wrote: > On Wednesday 08 September 2010 08:50:29 martin f krafft wrote: > [ /me thinks that we'll get similar problems witf config-edit-sshd, but I > digress ... ] > > By the way, what happens in your use case if the content of sshd_config is > *a

Bug#596038: Writes temporary files all over the filesystem

2010-09-14 Thread Dominique Dumont
On Wednesday 08 September 2010 08:50:29 martin f krafft wrote: > Augtool insists on writing to the filesystem, even if it does not > need to make any changes. For instance, the following fails because > /etc is mounted read-only, and hence /etc/ssh/sshd_config.augnew > cannot be written (strace att

Bug#596038: Writes temporary files all over the filesystem

2010-09-13 Thread Nicolas Valcárcel Scerpella
I don't understand the use case of this, why would you want to 'save' and that the changes aren't written to the disk? whouldn't that be unusefull? On Wed, 08 Sep 2010, martin f krafft wrote: > Package: augeas > Version: 0.7.0-1 > Severity: normal > Tags: upstream > > Augtool insists on writing

Bug#596038: Writes temporary files all over the filesystem

2010-09-07 Thread martin f krafft
Package: augeas Version: 0.7.0-1 Severity: normal Tags: upstream Augtool insists on writing to the filesystem, even if it does not need to make any changes. For instance, the following fails because /etc is mounted read-only, and hence /etc/ssh/sshd_config.augnew cannot be written (strace attached