Thunderbird and Firefox both fail to build due to failure similar to what is explained in this Mozilla bug report: https://bugzilla.mozilla.org/show_bug.cgi?id=732340 Fix: Attached patch taken verbatim from the bugzilla and expanded a bit to include explicit instantiation of template int SortedArrayOf<RangeRecord>::search<unsigned int>(unsigned int const &) const; Apparently, other people get other symbols undefined, so I left all four in. How-To-Repeat: Just try to build port using system GCC with no special options.
Responsible Changed From-To: freebsd-ports-bugs->gecko Over to maintainer.
State Changed From-To: open->closed A fix for this has been committed a couple of days ago. Please update your portstree and try to build them again. Thanks.
State Changed From-To: closed->open 'Fix build with base gcc by lowering the optimization level' This counts as a proper fix somehow?
Hi all. GCC 4.2.1 is rather a nice mess and known for generation of bad code with higher optimization levels. So yes, this looks quite logical for me. -- Sphinx of black quartz judge my vow.
On Mon, Jun 25, 2012 at 04:51:59PM +0300, Volodymyr Kostyrko wrote: > Hi all. > > GCC 4.2.1 is rather a nice mess and known for generation of bad code > with higher optimization levels. So yes, this looks quite logical for me. > > -- > Sphinx of black quartz judge my vow. Every compiler and compiler version has bugs. Should we make ports _never_ use anything more than -O just in case? After all, that was the official project stance on toolchain for a while. -- Alexander Kabaev
State Changed From-To: open->closed No point arguing.
State Changed From-To: closed->open Re-open.
State Changed From-To: open->closed Builds fine with base gcc again.