Summary: | editors/libreoffice: update 6.2.5 -> 6.3.0 compiler failure: no viable constructor ... | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | Ports & Packages | Reporter: | O. Hartmann <ohartmann> | ||||||
Component: | Individual Port(s) | Assignee: | Li-Wen Hsu <lwhsu> | ||||||
Status: | Closed FIXED | ||||||||
Severity: | Affects Many People | CC: | Wilbur310, adridg, dim, gja822, lwhsu, office, pi, vvd, w.schwarzenfeld, zirias | ||||||
Priority: | --- | Keywords: | regression | ||||||
Version: | Latest | Flags: | bugzilla:
maintainer-feedback?
(office) |
||||||
Hardware: | Any | ||||||||
OS: | Any | ||||||||
See Also: | https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=240064 | ||||||||
Attachments: |
|
Description
O. Hartmann
2019-08-20 12:34:13 UTC
(In reply to O. Hartmann from comment #0) Which version of FreeBSD are you using? I did test in pourdiere in 12.0 and 11.3, amd64 and i386. May I suggest you updating all the depends ports to the latest version? You can obtain the list via `make all-depends-list` I bet O.Harmann is on 13.0. (In reply to Walter Schwarzenfeld from comment #2) That should also be fine, my desktop is 13 and I updated it and test again right before committing. The failure happens on 13-CURRENT (r351229) and on 12-STABLE (last updated on Monday, revision at this moment not accessible due to off site). qt5 is per default off. 12.0-p9 amd64 and i386 - same error. (In reply to VVD from comment #6) OPTIONS_FILE_SET+=CUPS OPTIONS_FILE_SET+=DOCS OPTIONS_FILE_UNSET+=GNOME OPTIONS_FILE_UNSET+=GTK2 OPTIONS_FILE_UNSET+=GTK3 OPTIONS_FILE_SET+=JAVA OPTIONS_FILE_SET+=KDE5 OPTIONS_FILE_SET+=MMEDIA OPTIONS_FILE_UNSET+=PGSQL OPTIONS_FILE_SET+=QT5 OPTIONS_FILE_UNSET+=SDK OPTIONS_FILE_UNSET+=SYSTRAY OPTIONS_FILE_UNSET+=TEST OPTIONS_FILE_SET+=WEBDAV I can reproduce it by enabling QT5 option. Which revision of -CURRENT? Is that after base r351253? I'm on 12.0-RELEASE-p8 and facing the same issue, the build is broken with editors_libreoffice_UNSET+= GTK2 GTK3 editors_libreoffice_SET+= JAVA KDE5 SYSTRAY WEBDAV (probably caused by KDE5 option) (In reply to Felix Palmen from comment #10) > (probably caused by KDE5 option) KDE5_IMPLIES= QT5 Created attachment 206765 [details] removing using of std::function I haven't tried with base r351253, but this one looks like "Class template argument deduction for aggregates" feature, ref: https://clang.llvm.org/cxx_status.html https://bugs.llvm.org/show_bug.cgi?id=39606 Please try the attached workaround. Created attachment 206773 [details] removing using of std::function from vcl/unx/kde5/KDE5SalInstance.cxx (In reply to Li-Wen Hsu from comment #12) vcl/unx/kde5/KDE5SalInstance.cxx need the same patch. The same is on FreeBSD-11 STABLE (In reply to gja822 from comment #14) amd64 Compiled fine with our patches on 12.0 i386: https://bugs.freebsd.org/bugzilla/attachment.cgi?id=206765 https://bugs.freebsd.org/bugzilla/attachment.cgi?id=206773 A commit references this bug: Author: lwhsu Date: Thu Aug 22 06:21:32 UTC 2019 New revision: 509581 URL: https://svnweb.freebsd.org/changeset/ports/509581 Log: Workaround for build failure due to newer C++ features when KDE5 and QT5 defined PR: 239988 Submitted by: VVD <vvd@unislabs.com> Reported by: O. Hartmann <ohartmann@walstatt.org> Changes: head/editors/libreoffice/files/patch-vcl_qt5_Qt5Instance.cxx head/editors/libreoffice/files/patch-vcl_unx_kde5_kde5salinstance.cxx The latest commit fixes the build for my configuration, thanks! Please report this upstream (with the same fix; I've asked them to look at it already). It's a valid change, simplifies the code a bit. (In reply to Adriaan de Groot from comment #19) I'm still wondering and testing if my assumption in #12 is correct, I am not sure the feature used here is c++17 or c++2a. Also from the https://clang.llvm.org/cxx_status.html It seems that there are some features supported since clang7 and in current compiler.mk, it still uses clang6 for c++17. (In reply to Adriaan de Groot from comment #19) > Please report this upstream (with the same fix; I've asked them to look at it already). It's a valid change, simplifies the code a bit. https://bugs.documentfoundation.org/show_bug.cgi?id=127408 It was fixed upstream. MARKED AS SPAM |