Date: Sun, 13 Oct 2002 19:45:08 +0100 (BST) From: Sander Vesik <Sander.Vesik@Sun.COM> To: "Kevin B. Hendricks" <kevin.hendricks@sympatico.ca> Cc: dev@porting.openoffice.org, Martin Blapp <mb@imp.ch>, openoffice@FreeBSD.ORG Subject: Re: [porting-dev] Re: FreeBSD: mozilla datasource does not work Message-ID: <Pine.GSO.4.10.10210131932390.29425-100000@blossom> In-Reply-To: <200210131128.00608.kevin.hendricks@sympatico.ca>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 13 Oct 2002, Kevin B. Hendricks wrote: > Hi, > > I would be glad to add them but they will only work if the person who > checks them out has the same gcc version (for the same the C++ abi) as the > version that was checked in. > > So LinuxPPC would need to add versions for gcc 2.95.X, gcc 3.1.1, and gcc > 3.2 and even then there would be no guarantee that they would not need a > specific version of libgcc_so.1 and or libstdc++.so.X.Y.Z > > So right now on ppc linux, YellowDog would need one set, SuSE another, and > still a third for their upcoming releases. > > Perhaps we should only pick one specific gcc version to support for each > platform: > > linux - gcc 3.2.1 and glibc 2.3 since that is what everyone is moving to Is teher a need to use glibc 2.3 ? Unless soome radical changes happened (in which case it should really be called 3.0) a version compiled/linked against 2.1 should work on a 2.3 based system. If these were linked against the compilers libstdc++ (i think i know the cause why it presently might be an old libstdc++ version) then it would supposedly "just work" and eliminate an axis of difference we might otherwise have. > irix - gcc 2.95.X > freebsd - gcc 3.2.1 > MacOSX - gcc 3.1 (or we could pick gcc 2.95.X) depending on if we are > looking forward or backward > Yes, these are the largely simple cases of one known compiler we want to use. > etc. > > I just don't know the correct solution. Either way, without lots of moz > zip files lying around we can not guarantee buildability under linux for > gcc 2.95.X, 2.96, 3.0.4, gcc 3.1.X, and gcc 3.2.X which all seem to have > differnt abi's when it comes to C++ code. > I think we just need to make a pick - as we don't have a 2.96 bridge we obviously wouldn't include 2.96, for example. Would there be any actualy benefit from including 3.1.x version if everybody is moving to 3.2.x ? Also there is an additional hurdle in that the 1.0.x series needs tobe compatible so 1.0x on linux/x86 needs to remain using 3.0.x as the default compiler. Its a pity the '3.x.y will have a stable ABI' thing turned out to be thin air 8-( > Ideas anyone. > > Kevin > Sander There are voices in the street, And the sound of running feet, And they whisper the word -- Revolution! To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-openoffice" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.GSO.4.10.10210131932390.29425-100000>