Created attachment 861779 [details] gzip-compressed build.log The package dev-qt/qtdeclarative-5.15.9 fails to build due to apparent source code errors. The file gentoo-qconfig.h is empty. The build.log file is attached. emerge --info output: sprunge.us/zqYZrz
In future, please give emerge --info as a comment or attachment, as pastebins can expire.
Created attachment 861780 [details] emerge --info
By the way, I have tried re-emerging qtcore, qtgui, and qtnetwork.
There have been some reports with the config being empty due to, for example, something in the PATH variable making a find command fail [1][2], though there might be more causes than that [3] which remain unknown, and I don't see anything immediately odd in the 'emerge --info' provided PATH here at least... 1. https://forums.gentoo.org/viewtopic-t-1161660.html 2. https://forums.gentoo.org/viewtopic-t-1163101.html 3. https://forums.gentoo.org/viewtopic-t-1161250.html
(In reply to Chiitoo from comment #4) > There have been some reports with the config being empty due to, for > example, something in the PATH variable making a find command fail [1][2], > though there might be more causes than that [3] which remain unknown, and I > don't see anything immediately odd in the 'emerge --info' provided PATH here > at least... > > 1. https://forums.gentoo.org/viewtopic-t-1161660.html > 2. https://forums.gentoo.org/viewtopic-t-1163101.html > 3. https://forums.gentoo.org/viewtopic-t-1161250.html We need to, if nothing else, add a check in the eclass where it aborts if the thing it just generated is empty.
(In reply to Sam James from comment #5) > (In reply to Chiitoo from comment #4) > > There have been some reports with the config being empty due to, for > > example, something in the PATH variable making a find command fail [1][2], > > though there might be more causes than that [3] which remain unknown, and I > > don't see anything immediately odd in the 'emerge --info' provided PATH here > > at least... > > > > 1. https://forums.gentoo.org/viewtopic-t-1161660.html > > 2. https://forums.gentoo.org/viewtopic-t-1163101.html > > 3. https://forums.gentoo.org/viewtopic-t-1161250.html > > We need to, if nothing else, add a check in the eclass where it aborts if > the thing it just generated is empty. I think at least in one of the cases, it should have happened at around here: https://gitweb.gentoo.org/repo/gentoo.git/tree/eclass/qt5-build.eclass#n881 Which is a part of: # Don't call die here because dying in pkg_post{inst,rm} only makes things worse. qt5_regenerate_global_configs() { einfo "Regenerating gentoo-qconfig.h" find "${ROOT}${QT5_HEADERDIR}"/Gentoo \ -name '*-qconfig.h' -a \! -name 'gentoo-qconfig.h' -type f \ -execdir cat '{}' + | sort -u > "${T}"/gentoo-qconfig.h [[ -s ${T}/gentoo-qconfig.h ]] || ewarn "Generated gentoo-qconfig.h is empty" cp "${T}"/gentoo-qconfig.h "${ROOT}${QT5_HEADERDIR}"/Gentoo/gentoo-qconfig.h \ || eerror "Failed to install new gentoo-qconfig.h" einfo "Updating QT_CONFIG in qconfig.pri" So, there should at least be message(s) about it there, though no 'die' as it seems it might not be safe.
