On 2019-03-30, Andrew Savchenko <[email protected]> wrote: > On Sat, 30 Mar 2019 15:09:06 -0000 (UTC) Grant Edwards wrote: >> On 2019-03-29, Philip Webb <[email protected]> wrote: >> > 190329 Grant Edwards wrote: >> > >> >> gcc-7.3.9-r3 is marked stable, yet it fails to build if you have the >> >> current stable version of glibc installed (2.28-r5). >> > >> > I've been using Gcc-8.2.0-r6 since 170302 with Glibc-2.27-r6 : no problems. >> >> What I'm asking about is that 7.3.0-r3 (which is stable) won't build >> with glibc-2.28 (which is stable). My question: is that considered a >> bug or not? > > It depends on the details of the problem, but you provided no > details to make further considerations.
glibc 2.27 has an include file "ustat.h" which declares a library function ustat(). glibc 2.28 does not have that include file (nor the function, AFAICT). Any application that #includes ustat.h or calls ustat() fails to build with glibc 2.28. > In general it would be considered a bug. > >> One might think that the 7.3.0-r3 ebuild should require >> gblic < 2.28. Is one allowed to tweak ebuilds like that without >> bumping the revision? >> >> FWIW 7.3.0-r6 does build and works fine for my application which won't >> build with gcc-8 -- so it's purely an academic question. > > It's better to fix your application. Fixing problem revealed by gcc > update is usually not hard. My application uses a third-party open-source library for which I'm not a maintainer. That library contains code that can't be built with gcc 8.2. For various logistical reasons, I'm reluctant to make changes to that code. [It creates headaches and extra work down the road...] -- Grant
