Fog Creek Software
Discussion Board




Not a bug?

https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=119185

That is the funniest thing I have read this week.

I am not familiar with where this problem is. Is it in Linux itself or is it in RedHat?

Is Jeff a redhat employee or a Linux OSS dev? (Hopefully the latter, otherwise that argument of companies using RedHat "to get support" kinda goes out the window...)

Chris Ormerod
Friday, June 25, 2004

"Is Jeff a redhat employee or a Linux OSS dev? "

Seems more like a troll, we have plenty of them in this forum too.

Tom H
Friday, June 25, 2004

If he's a QA owner of rpm then he'll be an employee.

This proves my standard contention that only originators of bugs should be able to close them, not responders.

They might have a status of WONTFIX but it should still remain open.

Simon Lucy
Friday, June 25, 2004

Are you saying "Jeff" is a troll or me?

I was just pointing out something quite funny (in a developer sort of way), not meaning to troll for anything apart from maybe some opinions on wether the described functionality is by design or a bug.

Chris Ormerod
Friday, June 25, 2004

I think it's a kit-car special.  (BTW I only have limited Linux experince so feel free to correct me).

To start the bug appeared on Fedora Core - which is a sort of unsupported-development-use-at-your-own-risk -version of RedHat.  The relationship between FC & RH  is close.

RPM is the "Red Hat package manager" and is a widely used format across several distros.  YUM is the "Yellow Dog Updater (Modified)" which sits over the top of RPM asa sort of optional extra.

So there's a few questions to ask - is it down to RPM (which could effect a lot of distros) or YUM (which wouldn't) and if Jeff wossname does work for RH when did they start employing children.

a cynic writes...
Friday, June 25, 2004

a cynic writes....

Thats kinda what I was thinking, that Jeff guy can't be truly thinking that this isn't a bug unless he was a child. Quite clearly any normal program shouldn't be able to corrupt an important database just because a partition was readonly (and if it can corrupt a database that is stored on the readonly partition? that would make it an even worse bug - but I don't think this is the case.)

Chris Ormerod
Friday, June 25, 2004

To be honest, I think Jeff's caught in the wrong mental loop about the bug - for whatever reason he thought the bug was about RPM not being able to install to a read-only directory (which isn't a bug), and can't get his gears shifted. That's happened to me on occasion. I've learned when it seems that everyone is lost but me to go talk to a peer (or ask here) to try to figure out what I'm not getting.

It's like the Eudora loses mail bug - if you run out of hard drive space while Eudora is downloading mail, Eudora just dumps the lot. I'll bet some QA person reads the bug reports and thinks "well, what else can Eudora do when the disk is full?" without realizing the problem isn't pulling the mail to a full disk, it's deleting the mail from the server before guaranteeing delivery.

And yes, I'm sure MS has done its share of "misunderstanding the bug that's being reported" :-)

Philo

Philo
Friday, June 25, 2004

Hilarious.

muppet from madebymonkeys.net
Friday, June 25, 2004

"Are you saying "Jeff" is a troll or me?"

I was referring to Jeff, but I agree with Philo.

Basically Jeff said "of course it won't work, so don't do it ya moron". But he misses the point that if someone accidentally does do it, the sofware should handle it gracefully.

I've gotten the same response many times (usually sprinkled with lots of four letter words) over the years, it's everywhere.

Tom H
Friday, June 25, 2004

I've added comments to that effect and that a new bug should be opened.

It seems though that he's still stuck in the same loop.

I wonder if he's aware of the JOS effect?

Simon Lucy
Friday, June 25, 2004

Now I have read Philos response I understand how Jeff has been thinking.

I have done sort of the same thing, but never when the bug report was as detailed as this guys.

Chris Ormerod
Friday, June 25, 2004

Jeff seems to have closed the bug again. I love his argument that it is the users fault that their RPMDB is corrupt because they ran it while /usr was readonly.

Chris Ormerod
Friday, June 25, 2004

yeez, what an  a$$

i'd fire this guy, especially when the company is running on support income(like redhat does right?).

Guyon Morée
Friday, June 25, 2004

<troll>

Jeff sounds like a typical Linux developer to me....

</troll>

muppet from madebymonkeys.net
Friday, June 25, 2004

Wow, what a cool thread that was.

Even now, Jeff Johnson (whoever he is) just doesn't seem to 'get it', and he doesn't 'get' that he doesn't 'get it'.

Sure, the easiest answer is 'don't do that, it won't work'.  I'd like my car to fly, so I drive it off a cliff.  It breaks.  I complain it's a bug.  In that circumstance, 'don't DO that' is very appropriate.  (Also, it is not in the design of a car to fly.  At least, not MY car.)

I don't know, maybe Jeff analyzed how hard it would be to determine the /usr file-system was Read-Only, and then not do the things that corrupted the rmmdb database.  Maybe he concluded this was a lot of work, that would have little pay-off.  Maybe.

But I do agree with the thread posters.  Okay, so having '/usr' Read-Only is a fatal error to the install.  So be it.  But don't trash my rmmdb database in the process.  Add a check somewhere VERY early in the install process to see if '/usr' is Read-only, and abort before any damage is done. 

Preferably with a message saying "Oh!  By the way, your '/usr'  file system is Read-only.  Please change this and run again."

AllanL5
Friday, June 25, 2004

Simon--
Won't fix and not a bug are often different.
One means that it's a bug, but not important enough to fix.
The other means it's not a bug.
I don't know how this project manages its bugs, but a won't fix bug usually gets closed, to be possibly revisited later by looking at all closed 'won't fix' bugs.

mb
Friday, June 25, 2004

WONTFIX may mean that, but even if it does it shouldn't be used that way.

No bug is too trivial to fix, you may not schedule it for fixing but you shouldn't define it as WONTFIX, maybe some far off release tag.

Simon Lucy
Saturday, June 26, 2004

AllanL5:

I think the don't do that advice can be considered a workarround (in this situation)

I think your analogy is wrong. It should be more like: If I turn on my car radio and tune into a particular frequency, my speakers blow up.

Advice: Don't tune into that frequency
Bug: A wrong setting in a car radio shouldn't be able to blow up my speakers.

somemorone
Saturday, June 26, 2004

muppet: The fact that a half dozen other Linux developers jumped in to tell him what an idiot he's being seems to indicate that this isn't typical.  :-)

jim
Monday, June 28, 2004

*  Recent Topics

*  Fog Creek Home