Bug 192422 - rpmbuild doesn't produce debuginfos
rpmbuild doesn't produce debuginfos
Status: CLOSED NOTABUG
Product: Fedora
Classification: Fedora
Component: rpm (Show other bugs)
5
All Linux
medium Severity medium
: ---
: ---
Assigned To: Paul Nasrat
Mike McLean
: Reopened
Depends On:
Blocks:
  Show dependency treegraph
 
Reported: 2006-05-19 13:20 EDT by Ralf Corsepius
Modified: 2007-11-30 17:11 EST (History)
3 users (show)

See Also:
Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
Environment:
Last Closed: 2006-05-30 02:05:38 EDT
Type: ---
Regression: ---
Mount Type: ---
Documentation: ---
CRM:
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---


Attachments (Terms of Use)
src rpm exposing the problem (essentially hello world) (66.80 KB, application/x-redhat-package-manager)
2006-05-19 13:22 EDT, Ralf Corsepius
no flags Details

  None (edit)
Description Ralf Corsepius 2006-05-19 13:20:13 EDT
Description of problem:

rpmbuild doesn't build debuginfos, despite the package contains
binaries.

Also, on packages exposing these symptoms, executables don't seem get
stripped:
$ rpmlint ~/src/rpms/RPMS/i386/xxx-0.0.2-0.fc5.i386.rpm
...
W: xxx unstripped-binary-or-object /usr/bin/xyz

redhat-rpm-config is installed. The problem is reproducable in mock.

Version-Release number of selected component (if applicable):
redhat-rpm-config-8.0.40-1
rpm-build-4.4.2-15.2

How reproducible:
Only some rpms are affected. But for those which are, it seems to be deterministic.

Steps to Reproduce:
1. rpmbuild the rpm from the attachement.
2.
3.
  
Actual results:
No debuginfo.rpm being produced.

Expected results:
A debuginfo.rpm.

Additional info:
I haven't experienced this issue with Fedora < 5.
Comment 1 Ralf Corsepius 2006-05-19 13:22:28 EDT
Created attachment 129609 [details]
src rpm exposing the problem (essentially hello world)
Comment 2 Toshio Kuratomi 2006-05-19 13:50:40 EDT
For this package, adding an empty %build seems to solve the problem.
Comment 3 Jeff Johnson 2006-05-29 08:30:34 EDT
The redhat-rpm-config package overloads section markers within spec files to deposit the -debuginfo 
subpackage as a maggot. No section marker, no sub-package.

Hacky, isn't it?
Comment 4 Ralf Corsepius 2006-05-29 22:46:01 EDT
(In reply to comment #3)
> The redhat-rpm-config package overloads section markers within spec files
> to deposit the -debuginfo 
> subpackage as a maggot. No section marker, no sub-package.
>
> Hacky, isn't it?
Yes, crappy design, but why do you close this PR NOTABUG?

What else but "rpmbuild silently doesn't work as advertised" do you need to
qualify this as a bug?
Comment 5 Jeff Johnson 2006-05-30 02:05:38 EDT
NOTABUG because rpm is working exactly as requested:
    Produce -debuginfo with no change to spec files.

It's not an rpm problem that you are using a crappy and hacky design imperfectly.
Comment 6 Ralf Corsepius 2006-05-30 02:45:56 EDT
(In reply to comment #5)
> NOTABUG because rpm is working exactly as requested:
>     Produce -debuginfo with no change to spec files.
> 
> It's not an rpm problem that you are using a crappy and hacky design imperfectly.

OK, understood - you're never going to learn it. 

What does needs to happen until RH maintainers finally learn to do their job,
professionally?

Comment 7 Jeff Johnson 2006-05-30 11:33:35 EDT
Well, for starters, I'm an OSS, not a RH, maintainer these days, so I don't need the flame bait.
When are lusers like you gonna get a clue?
Comment 8 Ralf Corsepius 2006-05-30 12:02:39 EDT
(In reply to comment #7)
> Well, for starters, I'm an OSS, not a RH, maintainer these days, so I don't
> need the flame bait.
Well, I heard about it, but
1. I don't care for whom you work. 
2. I am complaining about you shipping broken software and closing bugs without
fixing them - I.e. I am accusing you to not doing your job as Fedora package
maintainer and am demanding consequences.


Comment 9 Michael Jennings (KainX) 2006-05-31 11:49:04 EDT
(In reply to comment #8)
> I am accusing you to not doing your job as Fedora package
> maintainer and am demanding consequences.

Demanding?  What gives you the right to demand anything?

It never ceases to amaze me that there are some members of the free software
community who not only are unsatisfied with getting something for nothing, but
rather they feel a sense of entitlement to get more for nothing then they're
already getting.

RPM is free in every sense of the word:  it's free to use, it's free to
modify/fix/enhance, and Jeff isn't getting paid to work on it.  So what exactly
is it that you think entitles you to make accusations or demands?  What is it
about you in particular that makes your opinion more correct or carry more
weight than Jeff's?

And just FYI, Jeff is not a Fedora package maintainer.  He's an upstream author.
 And I'd be willing to bet your demand of consequences gave him quite a good
chuckle.
Comment 10 Ralf Corsepius 2006-05-31 12:21:29 EDT
(In reply to comment #9)
> (In reply to comment #8)
> > I am accusing you to not doing your job as Fedora package
> > maintainer and am demanding consequences.
> 
> Demanding?  What gives you the right to demand anything?

My right as a user being nagged by broken SW.

> It never ceases to amaze me that there are some members of the free software
> community who not only are unsatisfied with getting something for nothing, but
> rather they feel a sense of entitlement to get more for nothing then they're
> already getting.
Let me put it this way: The OSS community is filled with blenders, polluting the
world with crappy SW and thereby being harmful to the OSS.

> RPM is free in every sense of the word:  it's free to use, it's free to
> modify/fix/enhance, and Jeff isn't getting paid to work on it.  So what exactly
> is it that you think entitles you to make accusations or demands?
> And just FYI, Jeff is not a Fedora package maintainer.  He's an upstream
> author.

Jeff closed my PR, without fixing it. He thereby acted as RH maintainer of rpm
and made it clear that he is refusing to acknowledge the bug. If he is not the
RH rpm package maintainer he should not have closed it.

I don't have a problem with bugs remaining documented as PRs and staying around
forever - I do have a problem with RH package maintainers actively closing PRs
on real bugs.

>  And I'd be willing to bet your demand of consequences gave him quite a good
> chuckle.
To me it's a riddle how it RH can tolerate to base their distros on a such
broken and unreliable piece of SW, being maintained the poor way it still is.
You might not be aware about it, but Jeff has a many years long record in
ignoring and closing bugs unfixed - If that RH's official bug handling policy,
do us all a favor and never ever ask users to file PRs on bugs.

Comment 11 Michael Jennings (KainX) 2006-05-31 15:43:43 EDT
(In reply to comment #10)
> My right as a user being nagged by broken SW.

You have only those rights granted you by the program's license.  The GPL and
the license terms supplied with RPM specifically state:

    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.

That means you are without any remedies whatsoever in the event you deem the
software "broken."  And, in fact, it is NOT broken.  It's just not behaving as
you expected, as Jeff quite clearly pointed out.

> ...bovine fecal matter omitted...
>
> To me it's a riddle how it RH can tolerate to base their distros on a such
> broken and unreliable piece of SW, being maintained the poor way it still is.
> You might not be aware about it, but Jeff has a many years long record in
> ignoring and closing bugs unfixed - If that RH's official bug handling policy,
> do us all a favor and never ever ask users to file PRs on bugs.

Your inability to comprehend an explanation or your refusal to accept it does
not constitute a bug.  If you feel RPM is so broken and unreliable, use
something else.

Jeff is quite honest and straight-forward when closing a bug in explaining what
the *real* issue is, and if applicable, why it can't or won't be fixed.  In my
experience, he's almost always right.  And he's right in this case.

The "debuginfo" mechanism was written to require a %build section and a %install
section as BOTH of these are customary for packages with binaries.  It makes
perfect sense.  If you don't have a %build, you don't have a binary; any other
circumstance is a bad spec file.

You don't have to like the fact that it was done that way, but it is not a bug.
 The simple fact is that you are improperly using an imperfect mechanism. 
Provide a better mechanism, or accept the word of the guy who knows a hell of a
lot more than you.

Either way, stop acting like the stereotypical whiny self-centered user and man up.
Comment 12 Ralf Corsepius 2006-05-31 18:26:22 EDT
(In reply to comment #11)
> (In reply to comment #10)
> > My right as a user being nagged by broken SW.
> 
> You have only those rights granted you by the program's license.  The GPL and
> the license terms supplied with RPM specifically state:
You are completely missing the point - What I said was: RH's management failed
to monitor rpm's development.

> And, in fact, it is NOT broken.  It's just not behaving as
> you expected, as Jeff quite clearly pointed out.
Oh yeah, it's not a bug it's a feature ... 

> > To me it's a riddle how it RH can tolerate to base their distros on a such
> > broken and unreliable piece of SW, being maintained the poor way it still is.
> > You might not be aware about it, but Jeff has a many years long record in
> > ignoring and closing bugs unfixed - If that RH's official bug handling policy,
> > do us all a favor and never ever ask users to file PRs on bugs.
> 
> Your inability to comprehend an explanation or your refusal to accept it does
> not constitute a bug.  If you feel RPM is so broken and unreliable, use
> something else.
Yes, RPM is one of the worst pieces of OSS SW around and RPM.specs are "so
broken it hurts" (citation JBJ). 
 
> The "debuginfo" mechanism was written to require a %build section and a %install
> section as BOTH of these are customary for packages with binaries.  It makes
> perfect sense.  If you don't have a %build, you don't have a binary; any other
> circumstance is a bad spec file.

Your assumption %build to be mandatory is wrong.

* Most configurations can be built by a simple "make install" from inside
%install, without a separate %build-stage.
* Inside of rpm.specs, %build is optional.

> You don't have to like the fact that it was done that way, but it is not
> a bug.
Oh yeah. You switch on building debug-infos but they don't get built, because
you haven't added %build? You switch on your car's lights but they don't ignite,
because you haven't switched on the air condition?

Do you feel the insanity of this?

> Either way, stop acting like the stereotypical whiny self-centered user
> and man up.
Yes, I stop arguing now. I don't see any sense in continuing this fruitless
discussions. You guys are simply keeping your eyes in front of the brokenness of
one of the foundations RH distros are built on.
Comment 13 Michael Jennings (KainX) 2006-06-01 09:15:11 EDT
(In reply to comment #12)
> You are completely missing the point - What I said was: RH's management failed
> to monitor rpm's development.

Not once did you say that.

> Oh yeah, it's not a bug it's a feature ... 

No, it's neither.  It's just the way debuginfo packages work.

> Yes, RPM is one of the worst pieces of OSS SW around and RPM.specs are "so
> broken it hurts" (citation JBJ). 

Good.  Stop using it.

> Your assumption %build to be mandatory is wrong.

See?  Not only have you failed to comprehend Jeff's explanation, but you've also
failed to comprehend most of what I've written.  Either there's a language
barrier here, or you simply can't or won't understand what we're saying.  But
for the benefit of others, I'll at least respond to your erroneous statements.

> * Most configurations can be built by a simple "make install" from inside
> %install, without a separate %build-stage.

There is a big difference between "can" and "should."  There are many, many
things you *can* do in a spec file which will get your package rejected by any
reasonable packaging standard.  Building binaries without a %build section would
be one of those things.

> * Inside of rpm.specs, %build is optional.

I never said it wasn't.  What I said was that having a spec which builds
binaries and lacks a %build is wrong.  As a matter of policy, not syntax.

> Oh yeah. You switch on building debug-infos but they don't get built, because
> you haven't added %build? You switch on your car's lights but they don't
> ignite, because you haven't switched on the air condition?

A more accurate analogy would be that you turn on the defroster and then try to
turn off the A/C, which doesn't work because the car was designed to activate
the A/C any time the defroster is turned on.  It may not fit your definition of
desired behavior, but it's part of the design, not a bug.

> Yes, I stop arguing now. I don't see any sense in continuing this fruitless
> discussions. You guys are simply keeping your eyes in front of the brokenness of
> one of the foundations RH distros are built on.

There are numerous distributions out there which are not based on RPM.  Perhaps
you'd feel more comfortable with one of those instead.

Note You need to log in before you can comment on or make changes to this bug.