SYSTEMD: Give us a option for upstart

Jeff Spaleta jspaleta at gmail.com
Tue Jun 14 15:12:15 UTC 2011


On Tue, Jun 14, 2011 at 6:32 AM, Orcan Ogetbil <oget.fedora at gmail.com> wrote:
> I understand the inconsistency and it is indeed a bug in mount.


>
> Nevertheless you are missing the point. If X worked before (X=mounting
> at boot with fstab containing trailing slashes), and stops working now
> because of the change Y I made,

I can't remember seeing ever seeing an fstab that uses trailing
slashes in an operational system that I've had access to.  We don't
auto-generate trailing slash mount points entries in fstab. I didn't
even _think_ to test it when I was doing my due diligence for systemd
testing during pre-release.  And since I know I'm smarter than
everyone involved with systemd development put together, it would be
unfair of me to expect them to have caught this

So why have I never seen a trailing slash on an operational fstab
mount point on a linux system? Why don't we generate trailing slash
mountpoints automatically in our default fstab config on install?
Most likely, because experienced sysadmins over the years have
probably conditioned themselves to avoid using trailing slash entries
because of mount's existing cmdline quirk and everyone's been too
busy/lazy to file the bug to get mount fixed. Until you run into it on
the cmdline yourself, its not noticable. And even then its easier to
just fix your custom mountpoints and remove the slash.


> I am responsible for fixing X or Y.
Fixed.The utils-linux developers have fixed it very quickly once
someone actually filed a bug.

> The question of 'which one contains the bug' is irrelevant for the
> user.

Sure...I'm not saying otherwise. But this isn't a user list. This is a
devel list and we are having a discussion about development.  From a
user perspective it just needs to get fixed...and it is going to get
fixed with an updated utils-linux as soon as possible. From a strict
user perspective problem solved.

> Some folks think that this is a corner case and it is easy to
> miss.

I didn't think to test it. Did you test it? There is no evidence
what-so-ever that anyone actually tested this prior to release. And as
far as I know the person who ran into this is the only person on the
planet who writes trailing slash fstab entries.  And since we..in
Fedora...don't populate fstab entries by default in the install or
live images with trailing slashes there's no _expectation_ that this
will be tested as part of QA testing.  I'm going to re-iterate that.
QA has finite resources and a narrowly defined testing mandate.
Syntax quirks of this nature which are not used in the install targets
will not be found prior to release without the help of people out in
the userbase who are willing to volunteer and test their real world
setups which diverge from the default configs.

If trailing slashes were so common as to not be thought of as a
corner-case then it is reasonable to assume someone would have hit
this prior to release and shouted about it on one of the lists. Didn't
happen.

> I think that this is a fundamental mistake and this should be
> one of the first things a programmer should learn. It pretty much
> compares a physicist forgetting F=ma. Well, we all do mistakes.

Speaking as a PhD physicist who develops and maintains software for
experimental research that an international collaboration of PhD
physicists and students blindly rely on in order to do science without
being expected to understanding how the actual hardware or software
works in detail... I think you are very wrong.
Simply put, F=ma is well documented. Mount's slash handling behavior
is undocumented. If its undocumented behavior there is no expectation
that it can be tested or verified.

> Unfortunately, it caused problems for at least a couple people.
> Hopefully the programmer learned his lesson.

Sure a couple of people, who did not invest in the pre-release testing
process. I hope those couple of people learned their lesson.

-jef


More information about the devel mailing list