Skip to contentSkip to conversation

m4-mative: "Please port gnulib fseeko.c to your platform!"

DISCUSSION Robert P. J. Day 2 participants 6 messages Apr 8, 2019 – Apr 9, 2019

lore
Share & tools
lore
#0 lore raw

a few months ago, i ran into (and asked about) a failure building
m4-native that you can read here:

https://yocto.yoctoproject.narkive.com/6KyM5UvS/m4-native-zedboard-please-port-gnulib-fseeko-c-to-your-platform

first, i will admit i'm not sure where the fault lies -- is it m4, or gnulib, or glibc, and whose responsibility is it to fix what. i'll be honest -- i don't understand what the bug is or who should be resolving it.

i ask all this as i just had *exactly* the same error trying to
build agl-image-minimal with the AGL code base. what is the proper solution to this? BTW, i'm trying this on fedora 30 branched, so this is definitely not a supported build distro, but i'm still curious as to what would be the proper fix.

finally, it occurs to me that, if building a native package like
"m4" fails, would it be a reasonable workaround to simply say

ASSUME_PROVIDED += "m4"

for the time being, to avoid having to build it until i know what the proper fix is?

rday

Signature · 7 lines
========================================================================
Robert P. J. Day                                 Ottawa, Ontario, CANADA
                         http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================
#1 lore raw
Quoted 17 lines · Robert P. J. Day

a few months ago, i ran into (and asked about) a failure building
m4-native that you can read here:

https://yocto.yoctoproject.narkive.com/6KyM5UvS/m4-native-zedboard-please-port-gnulib-fseeko-c-to-your-platform

first, i will admit i'm not sure where the fault lies -- is it m4, or gnulib, or glibc, and whose responsibility is it to fix what. i'll be honest -- i don't understand what the bug is or who should be resolving it.

i ask all this as i just had *exactly* the same error trying to
build agl-image-minimal with the AGL code base. what is the proper solution to this? BTW, i'm trying this on fedora 30 branched, so this is definitely not a supported build distro, but i'm still curious as to what would be the proper fix.

Quoted 7 lines

finally, it occurs to me that, if building a native package like
"m4" fails, would it be a reasonable workaround to simply say

ASSUME_PROVIDED += "m4"

for the time being, to avoid having to build it until i know what the proper fix is?

trying ASSUME_PROVIDED didn't work all that well, either ...

...
| autom4te: need GNU m4 1.4 or later: m4
| make[2]: *** [Makefile:641: autoconf.in] Error 1
...

that's just weird as my installed version of m4 is 1.4.18, precisely
the version AGL is trying to build. i'm confused.

rday

Signature · 7 lines
========================================================================
Robert P. J. Day                                 Ottawa, Ontario, CANADA
                         http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================
#2 lore raw
In reply to Robert P. J. Day

Looks like that bug in gnulib that broke m4 (and others) when glibc changed.

Fixed upstream https://git.savannah.gnu.org/cgit/gnulib.git/commit/?id=4af4a4a718
and oe-core master/thud/possibly sumo also carry the relevant patch for m4.

If you're using older releases against a newer release then 1) this is why we have supported host distros and 2) the patch is easy to backport.

Ross

Quoted 42 lines · Robert P. J. Day

a few months ago, i ran into (and asked about) a failure building
m4-native that you can read here:

https://yocto.yoctoproject.narkive.com/6KyM5UvS/m4-native-zedboard-please-port-gnulib-fseeko-c-to-your-platform

first, i will admit i'm not sure where the fault lies -- is it m4, or gnulib, or glibc, and whose responsibility is it to fix what. i'll be honest -- i don't understand what the bug is or who should be resolving it.

i ask all this as i just had *exactly* the same error trying to
build agl-image-minimal with the AGL code base. what is the proper solution to this? BTW, i'm trying this on fedora 30 branched, so this is definitely not a supported build distro, but i'm still curious as to what would be the proper fix.

finally, it occurs to me that, if building a native package like
"m4" fails, would it be a reasonable workaround to simply say

ASSUME_PROVIDED += "m4"

for the time being, to avoid having to build it until i know what the proper fix is?

rday

--

========================================================================
Robert P. J. Day Ottawa, Ontario, CANADA
http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday

========================================================================
--
_______________________________________________
yocto mailing list
[email protected]
https://lists.yoctoproject.org/listinfo/yocto

#3 lore raw
Quoted 11 lines · Ross

Looks like that bug in gnulib that broke m4 (and others) when glibc changed.

Fixed upstream
https://git.savannah.gnu.org/cgit/gnulib.git/commit/?id=4af4a4a718
and oe-core master/thud/possibly sumo also carry the relevant patch for m4.

If you're using older releases against a newer release then 1) this is why we have supported host distros and 2) the patch is easy to backport.

so, to confirm what i thought, this is a bug in gnulib which
suggests i should dig through the build configuration to see where i can, say, set a PREFERRED_VERSION of a newer version of gnulib that no longer has tat bug? or, barring that, just add the appropriate local patch on my end.

rday

Signature · 7 lines
========================================================================
Robert P. J. Day                                 Ottawa, Ontario, CANADA
                         http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================
#4 lore raw
Quoted 5 lines · Robert P. J. Day

so, to confirm what i thought, this is a bug in gnulib which
suggests i should dig through the build configuration to see where i can, say, set a PREFERRED_VERSION of a newer version of gnulib that no longer has tat bug? or, barring that, just add the appropriate local patch on my end.

No, gnulib is a copy/paste library. The bug was fixed in gnulib, and new releases of m4 are fixed, but you'll need to cherry-pick the fix from oe-core master (we apply the patch to our m4 recipe) back to whatever stable branch you're using.

Ross

#5 lore raw
Quoted 11 lines · Ross
Quoted 5 lines · Robert P. J. Day

so, to confirm what i thought, this is a bug in gnulib which
suggests i should dig through the build configuration to see where i can, say, set a PREFERRED_VERSION of a newer version of gnulib that no longer has tat bug? or, barring that, just add the appropriate local patch on my end.

No, gnulib is a copy/paste library. The bug was fixed in gnulib, and new releases of m4 are fixed, but you'll need to cherry-pick the fix from oe-core master (we apply the patch to our m4 recipe) back to whatever stable branch you're using.

ah, got it, thanks.

rday