Bug 1421088 - Build succeeds even if the make inside fails
Summary: Build succeeds even if the make inside fails
Keywords:
Status: CLOSED RAWHIDE
Alias: None
Product: Fedora
Classification: Fedora
Component: snappy-java
Version: rawhide
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Pavel Raiskup
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2017-02-10 10:20 UTC by Tomas Repik
Modified: 2017-02-10 15:53 UTC (History)
4 users (show)

Fixed In Version: 1.1.2.4-4.fc26
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2017-02-10 15:53:44 UTC
Type: Bug


Attachments (Terms of Use)
cause error in make (684 bytes, patch)
2017-02-10 10:20 UTC, Tomas Repik
no flags Details | Diff
Proposed fix. (1.36 KB, patch)
2017-02-10 15:22 UTC, Pavel Raiskup
no flags Details | Diff

Description Tomas Repik 2017-02-10 10:20:04 UTC
Created attachment 1249012 [details]
cause error in make

Description of problem:
The build of snappy-java goes successfuly true even if the make ends in error.
I guess it's not what should happen.


Version-Release number of selected component (if applicable):
snappy-java-1.1.2.4-3

How reproducible:
Make the make command fail, for example by adding a syntax error to  src/main/java/org/xerial/snappy/SnappyNative.cpp.

Steps to Reproduce:
build snappy-java with attached patch

Actual results:
build succeed

Expected results:
build fails

Additional info:
https://kojipkgs.fedoraproject.org//work/tasks/8026/17708026/build.log

Comment 1 gil cattaneo 2017-02-10 10:33:48 UTC
i try to found the syntax error in src/main/java/org/xerial/snappy/SnappyNative.cpp but in our source archive is not available.
(contets of SnappyNative.cpp)
#include <string>
#include <cstring>
#include <snappy.h>
#include "SnappyNative.h"

where you download/found this?

Comment 2 Tomas Repik 2017-02-10 11:22:03 UTC
I'm not saying there is a syntax error in the sources. I'm saying that if there is an error one the build should fail.

I made the error on purpose to illustrate the behavior.

Comment 3 gil cattaneo 2017-02-10 12:01:40 UTC
(In reply to Tomas Repik from comment #2)
> I'm not saying there is a syntax error in the sources. I'm saying that if
> there is an error one the build should fail.
> 
> I made the error on purpose to illustrate the behavior.

Default build tool for this project is sbt (= 0.13.8), and a lot of sbt plugins,  not available in our repositories. So we use maven. I/we do MANY CAUTION during the build process to AVOID problem with COMPILATION. For me this bug is not valid

Comment 4 gil cattaneo 2017-02-10 12:05:00 UTC
Latest scratch build Task info: https://koji.fedoraproject.org/koji/taskinfo?taskID=17709474

Comment 5 Pavel Raiskup 2017-02-10 12:37:56 UTC
This _is_ a bug, Gil.  If for example, for whatever reason, g++ refuses to
build the only *.cpp file in future -- nobody will notice (koji build
succeeds, so what?).  That's like you've done '%mvn_build || true' to hide
the errors..  This would be a bit similar bug 1420790 .. it survived one
stable release broken.

If you claim this is not a bug in snappy-java, please switch against
appropriate component, it would be really helpful!

If you have not enough time to fix yourself, can you please guide us
towards correct solution?  What we need is:  to ensure that when the
compilation fails, also the whole koji build fails so we can observe FTBFS
as soon as possible.

Comment 6 Pavel Raiskup 2017-02-10 12:48:35 UTC
(In reply to Pavel Raiskup from comment #5)
> This _is_ a bug, Gil.  If for example, for whatever reason, g++ refuses to
> build the only *.cpp file in future -- nobody will notice

I admit that if you do the build, you will notice most probably, but
for example Fedora release engineers might spin build during mass rebuild
and broken package could get into repository.

Feel free to assign this bug to me or Tomáš, because we plan to have this
fixed, but yeah - help would be appreciated.

Comment 7 gil cattaneo 2017-02-10 12:52:19 UTC
(In reply to Pavel Raiskup from comment #5)
> This _is_ a bug, Gil.  If for example, for whatever reason, g++ refuses to
> build the only *.cpp file in future -- nobody will notice (koji build
> succeeds, so what?).  That's like you've done '%mvn_build || true' to hide
> the errors..  This would be a bit similar bug 1420790 .. it survived one
> stable release broken.

This is only an your opinion and not involved me and the package

> If you claim this is not a bug in snappy-java, please switch against
> appropriate component, it would be really helpful!
I dont care
> If you have not enough time to fix yourself, can you please guide us
> towards correct solution?  What we need is:  to ensure that when the
> compilation fails, also the whole koji build fails so we can observe FTBFS
> as soon as possible.

No solution. The only one is take care of the build process

Comment 8 gil cattaneo 2017-02-10 12:56:22 UTC
(In reply to Pavel Raiskup from comment #6)
> (In reply to Pavel Raiskup from comment #5)
> > This _is_ a bug, Gil.  If for example, for whatever reason, g++ refuses to
> > build the only *.cpp file in future -- nobody will notice
> 
> I admit that if you do the build, you will notice most probably, but
> for example Fedora release engineers might spin build during mass rebuild
> and broken package could get into repository.

If something in the build process gone wrong. we can notice its. This library is used in many projects as build or test deps, using Koschei.

> Feel free to assign this bug to me or Tomáš, because we plan to have this
> fixed, but yeah - help would be appreciated.

I will close this bug for the last time.

Comment 9 Pavel Raiskup 2017-02-10 13:22:59 UTC
> No solution. The only one is take care of the build process

Which means manual review of all of this?  If yes then -1:
https://koji.fedoraproject.org/koji/packageinfo?packageID=13533

> If something in the build process gone wrong. we can notice its. This
> library is used in many projects as build or test deps, using Koschei.

-1.  Without doubts, Koschei is too late for such build failures.  Please
don't bother other package maintainers with FTBFS in package you maintain.

> I will close this bug for the last time.

Uh, OK :(.  Thanks anyway for being responsive and sorry for me being that
unyielding.  I'm reopening again and assigning to myself, I hope I'll have
a time to have a look soon ... I'll propose a fix or close CANTFIX.  Still
any hint is appreciated.

Comment 10 Pavel Raiskup 2017-02-10 15:22:08 UTC
Created attachment 1249060 [details]
Proposed fix.

Gil, can you please have a quick look?  The patch is in private branch
private-praiskup-rhbz-1421088.  I can merge && build this in F26.

Comment 11 gil cattaneo 2017-02-10 15:24:42 UTC
i can apply the patch now

Comment 12 Pavel Raiskup 2017-02-10 15:27:44 UTC
Thanks!

Comment 13 gil cattaneo 2017-02-10 15:29:53 UTC
Task info: https://koji.fedoraproject.org/koji/taskinfo?taskID=17722119


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