Installation problem: Manjaro and libkd

[rmgreve]

Hello everybody and a happy new year!

I have got the following problem during the installation: System: Manjaro, x86_64 Linux i7 4.14.160-1-MANJARO #1 SMP PREEMPT

I will try: 1) yaourt -Ss aur/astrometry.net 0.78-3 (5) (0,00) Automatic recognition of astronomical images

but 2) yaourt -S astrometry.net

i will get the info:

at the beginning:

==> Erstelle und installiere Paket ==> Erstelle Paket: astrometry.net 0.78-3 (Do 02 Jan 2020 07:49:26 CET) ==> Prüfe Laufzeit-Abhängigkeiten... ==> Prüfe Buildtime-Abhängigkeiten... ==> Empfange Quellen... -> Lade astrometry.net-0.78.tar.gz herunter... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 10.9M 100 10.9M 0 0 4173k 0 0:00:02 0:00:02 --:--:-- 4171k ==> Überprüfe source Dateien mit md5sums... astrometry.net-0.78.tar.gz ... Durchgelaufen ==> Entpacke Quellen... -> Entpacke astrometry.net-0.78.tar.gz mit bsdtar ==> Beginne prepare()... ==> Beginne build()... make -C util config make -C qfits-an make -C sdss echo "version = '0.78'" > init.py make -C libkd pyspherematch make -C util cairoutils.o make[1]: Verzeichnis „/tmp/yaourt-tmp-rmgreve/aur-astrometry.net/src/astrometry.net-0.78/qfits-an“ wird betreten make[1]: Verzeichnis „/tmp/yaourt-tmp-rmgreve/aur-astrometry.net/src/astrometry.net-0.78/util“ wird betreten make[1]: Verzeichnis „/tmp/yaourt-tmp-rmgreve/aur-astrometry.net/src/astrometry.net-0.78/sdss“ wird betreten make[1]: Verzeichnis „/tmp/yaourt-tmp-rmgreve/aur-astrometry.net/src/astrometry.net-0.78/libkd“ wird betreten make[1]: Verzeichnis „/tmp/yaourt-tmp-rmgreve/aur-astrometry.net/src/astrometry.net-0.78/util“ wird betreten

####################################################################################### at the end:

make[1]: Verzeichnis „/tmp/yaourt-tmp-rmgreve/aur-astrometry.net/src/astrometry.net-0.78/util“ wird verlassen make[1]: Verzeichnis „/tmp/yaourt-tmp-rmgreve/aur-astrometry.net/src/astrometry.net-0.78/libkd“ wird verlassen make: *** [Makefile:116: libkd-spherematch] Fehler 2 ==> FEHLER: Ein Fehler geschah in build(). Breche ab... ==> FEHLER:Makepkg konnte astrometry.net nicht erstellen. ==> Erstellen von astrometry.net neu starten?[j/N]

It looks like libkd-spherematch is the problem...

What is going wrong?

Thank you very much !

Matthias

[dstndstn]

You need to show more of the error. Please send the whole log.

[pozar87]

Hi! I have a similar (if not the same) issue. Here are logs form my build. Best Regards!

[dstndstn]

Is it trying to do a parallel build? The Astrometry.net makefiles do NOT work reliably with "make -j8" etc

[pozar87]

Hi thanks for a quick answer! It doesn't seem so - the full script used by Arch/Manjaro to build the package is available here: https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=astrometry.net I've tried pass -j1 to the make command but the problem persists.

[dstndstn]

You are certain of that? In your build.log I see:

""" ---- Error messages in the next few commands are not necessarily bugs ---- (we're checking how things works on your computer) ---- Error messages in the next few commands are not necessarily bugs ---- ... (we're checking how things works on your computer) """

which looks like two processes both trying to build os-features-config.log , and then

""" --------------- End of expected error messages -----------------

mv ../include/astrometry/os-features-config.h.tmp ../include/astrometry/os-features-config.h

...

mv ../include/astrometry/os-features-config.h.tmp ../include/astrometry/os-features-config.h mv: nie można wykonać stat na '../include/astrometry/os-features-config.h.tmp': Nie ma takiego pliku ani katalogu """

which definitely looks like two processes racing to rename the output file.

Maybe MAKEFLAGS is getting set?

cheers, --dustin

On Sat, Feb 29, 2020 at 5:50 AM Łukasz Pożarlik notifications@github.com wrote:

Hi thanks for quick answer! It doesn't seem so - the full script used by Arch/Manjaro to build the package is available here: https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=astrometry.net I've tried pass -j1 to the make command but the problem persists.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/dstndstn/astrometry.net/issues/178?email_source=notifications&email_token=AAIEH7L4TOBTHT7O7TOB4STRFDUBHA5CNFSM4KB7M42KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOENLW2NY#issuecomment-592932151, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAIEH7OLIN4IT5RVNEQBJI3RFDUBHANCNFSM4KB7M42A .

[pozar87]

Indeed! I valuate closely when and how MAKEFLAGS is set in Manjaro and found that default form /etc/makepkg.conf is set to number of processors + 1

MAKEFLAGS="-j$(($(nproc)+1))"

So I added export MAKEFLAGS="-j 1" to build file as a workaround and retry building. Error has changed and now its complaining about qsort_r as well as redefinition of HAVE_NETPBM: build.log

[dstndstn]

No, look at the log, it's still building in parallel.

make -C util config make -C qfits-an make -C sdss echo "version = '0.79'" > init.py make -C libkd pyspherematch make[1]: Wejście do katalogu '/home/pozar/.cache/yay/ astrometry.net/src/astrometry.net-0.79/util' make[1]: Wejście do katalogu '/home/pozar/.cache/yay/ astrometry.net/src/astrometry.net-0.79/qfits-an' make[1]: Wejście do katalogu '/home/pozar/.cache/yay/ astrometry.net/src/astrometry.net-0.79/sdss'

On Sun, Mar 1, 2020 at 6:34 AM Łukasz Pożarlik notifications@github.com wrote:

Indeed! I valuate closely when and how MAKEFLAGS is set in Manjaro and found that default form /etc/makepkg.conf is set to number of processors

• 1

MAKEFLAGS="-j$(($(nproc)+1))"

So I added export MAKEFLAGS="-j 1" to build file as a workaround and retry building. Error has changed and now its complaining about qsort_r as well as redefinition of HAVE_NETPBM: build.log https://github.com/dstndstn/astrometry.net/files/4271715/build.log

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/dstndstn/astrometry.net/issues/178?email_source=notifications&email_token=AAIEH7IE4KTYC7N5XWDJP33RFJB4PA5CNFSM4KB7M42KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOENM4TKA#issuecomment-593086888, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAIEH7IP4FZGAL2W7S7SVC3RFJB4PANCNFSM4KB7M42A .

[pozar87]

You are right again - there are two make commend invocations in the PKGBUILD and I missed one of them... With -j 1 it builds like a charm. I'll let maintainer of astrometry.net Arch package know that single-threaded build is required. Thanks for your support!

Feel free to close this issue :)

[dstndstn]

Thank you!

On Tue, Mar 3, 2020 at 5:02 AM Łukasz Pożarlik notifications@github.com wrote:

You are right again - there are two make commend invocations in the PKGBUILD and I missed one of them... With -j 1 it builds like a charm. I'll let maintainer of astrometry.net http://astrometry.net Arch package know that single-threaded build is required. Thanks for your support!

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/dstndstn/astrometry.net/issues/178?email_source=notifications&email_token=AAIEH7M3NGC2SPG6LZIM7Z3RFTIU3A5CNFSM4KB7M42KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOENS3CMQ#issuecomment-593867058, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAIEH7M5FJN6G37UPOBXPHDRFTIU3ANCNFSM4KB7M42A .

[oldherl]

Is there a plan to support parallel build?

[dstndstn]

my plan is (1) find someone who cares (2) get them to fix it :)