Re: GCC 3.2 patch
Quoting Alexander Kabaev [EMAIL PROTECTED]: | Hi everyone, | | I've collected a number of patches for several problems with | GCC 3.2 compiler which have been brought to my attention so far. | While I am waiting for these patches or other suitable fixes to be | incorporated into FSF CVS repository, I decided to make a patch | file available at http://people.freebsd.org/~kan/gcc-all.diff. | | People having problems compiling their problems with the new | system compiler are encouraged to give this patch a try and | let me know if their problem is fixed ot not. | | -- Alexander Kabaev | | To Unsubscribe: send mail to [EMAIL PROTECTED] | with unsubscribe freebsd-ports in the body of the message -- To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2 patch
Quoting Alexander Kabaev [EMAIL PROTECTED]: | Hi everyone, | | I've collected a number of patches for several problems with | GCC 3.2 compiler which have been brought to my attention so far. | While I am waiting for these patches or other suitable fixes to be | incorporated into FSF CVS repository, I decided to make a patch | file available at http://people.freebsd.org/~kan/gcc-all.diff. | | People having problems compiling their problems with the new | system compiler are encouraged to give this patch a try and | let me know if their problem is fixed ot not. | | -- Alexander, It did fix my problem. I did a cvsup and buildworld this morning over yesterday afternoon's application and build. It didn't even dawn on me until your email. I reapplied the patch and now it is fine. Thanks. I'll do more testing tomorrow but I'm sure it will solve some of my problems with port rebuilding. ed To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2 patch
On Fri, Sep 06, 2002 at 06:53:18PM -0700, Alexander Kabaev wrote: Hi everyone, I've collected a number of patches for several problems with GCC 3.2 compiler which have been brought to my attention so far. While I am waiting for these patches or other suitable fixes to be incorporated into FSF CVS repository, I decided to make a patch file available at http://people.freebsd.org/~kan/gcc-all.diff. People having problems compiling their problems with the new system compiler are encouraged to give this patch a try and let me know if their problem is fixed ot not. I'll test this on bento ASAP. Thanks! Kris msg42705/pgp0.pgp Description: PGP signature
Re: GCC 3.2
On Thu, Aug 15, 2002 at 03:47:47PM -0700, Terry Lambert wrote: And we all know how successful that was, right? On the other side, we all know how successfull we were trying to get GCC 2.95.x bugs fixed for us, right? Do you really want to repeat this deeply satisfying experiment again? That was because the patches were not being submitted back against the unadulterated distribution code someone who had signed the assignment of rights to the FSF. Blah Terry, TOTALLY 110% INCORRECT. The situation was the same as our FreeBSD 3.x users that still post PR's against RELENG_3 and want us to fix things. Even where there was complete patches against 2.94.3 available; the issue for the GCC people was one of not willing to spend the effort to re-test on all platforms. Same reason we don't upgrade RELENG_3 to the latest openssl (or any other lib) -- who knows what else would break that depended on version that is there now. The inability to get patches into 2.95 is totally unrelated to the fact that it was an older GCC, and completely related to the fact that the patches were not submitted in accordance with the GCC maintainer's guidelines, WRONG, WRONG, WRONG, WRONG !!! To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
David O'Brien wrote: And we all know how successful that was, right? On the other side, we all know how successfull we were trying to get GCC 2.95.x bugs fixed for us, right? Do you really want to repeat this deeply satisfying experiment again? That was because the patches were not being submitted back against the unadulterated distribution code someone who had signed the assignment of rights to the FSF. Blah Terry, TOTALLY 110% INCORRECT. The situation was the same as our FreeBSD 3.x users that still post PR's against RELENG_3 and want us to fix things. Even where there was complete patches against 2.94.3 available; the issue for the GCC people was one of not willing to spend the effort to re-test on all platforms. Same reason we don't upgrade RELENG_3 to the latest openssl (or any other lib) -- who knows what else would break that depended on version that is there now. I thought that this was true for the LD, but not true for the GCC. I think this is a different problem here, since this was a specific reference to GCC 2.95. I definitely agree that this was an issue for the linker; the 2.95 was, I thought, never that much out of date, at the time the FreeBSD specific patches were initially made. The inability to get patches into 2.95 is totally unrelated to the fact that it was an older GCC, and completely related to the fact that the patches were not submitted in accordance with the GCC maintainer's guidelines, WRONG, WRONG, WRONG, WRONG !!! It *was* an older GCC?!? Now I'm confused. We *are* talking about the a.out shared library support, right? -- Terry To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
On Mon, Aug 19, 2002 at 01:04:55PM -0700, Terry Lambert wrote: Blah Terry, TOTALLY 110% INCORRECT. The situation was the same as our FreeBSD 3.x users that still post PR's against RELENG_3 and want us to fix things. Even where there was complete patches against 2.94.3 available; the issue for the GCC people was one of not willing to spend the effort to re-test on all platforms. Same reason we don't upgrade RELENG_3 to the latest openssl (or any other lib) -- who knows what else would break that depended on version that is there now. I thought that this was true for the LD, but not true for the GCC. I think this is a different problem here, since this was a specific reference to GCC 2.95. It is more true for GCC than anything I maintain(ed) in src/contrib/ The inability to get patches into 2.95 is totally unrelated to the fact that it was an older GCC, and completely related to the fact that the patches were not submitted in accordance with the GCC maintainer's guidelines, WRONG, WRONG, WRONG, WRONG !!! It *was* an older GCC?!? Now I'm confused. We *are* talking about the a.out shared library support, right? Nope. We are talking about various exception and code generation bugs. ELF format and sjlj method. Very mainstream things for FreeBSD. To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
On Sun, Aug 18, 2002 at 03:27:31PM +0200, Martin Blapp wrote: Hi, Any plans or ideas when gcc3.2 will be imported ? Martin I think if you search the mailinglist archive you will find your answer quickly (it has been addressed several times). -- Morten Rodal // // PGP ID 2D75595B // 22DE D67A 1AEA EF94 872A 9384 6D67 B50B 2D75 595B // msg41985/pgp0.pgp Description: PGP signature
Re: GCC 3.2
Hi, I think if you search the mailinglist archive you will find your answer quickly (it has been addressed several times). Thanks, yes found it. But with the answers I'm very unpleased. I really really hope that we import either 3.2 or 3.3 now. Personally I'd go with 3.2. The fact is that several ports need at least gcc3.1.1. We still have a prerelease 3.1 with many bugs there. The situation is very unpleasant. Martin To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
mb The situation is very unpleasant. IIRC, we have no active GCC maintainer, no matter you feel unpleasant or not... -- - Makoto `MAR' Matsushita To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
According to Terry Lambert: There's always waiting for 3.3 to be released before trying to incorporate it... There are too many code generation bugs in our version right now. Some ports need 3.1.1 from ports (remember our gcc is 3.1-prerelease). I don't care about 3.2 or 3.3, but I'd say go for snap of 3.3 now, if you look at the ports gcc, gcc32 == gcc33 at the moment. -- Ollivier ROBERT -=- FreeBSD: The Power to Serve! -=- [EMAIL PROTECTED] FreeBSD keltia.freenix.fr 5.0-CURRENT #80: Sun Jun 4 22:44:19 CEST 2000 To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
RE: GCC 3.2
Hi, Are any plans to move to GCC 3.2 in current? Since it is just an ABI change it should work, without changing anything. It would give us a stable, multivendor ABI to work off of for the next line of 5.x releases. Just a thought. Jesse Gross Yes, moving to gcc32 is highly desirable for -current, otherwise we will be stuck at gcc311 for the entire life of FreeBSD 5.x. The important question to ask is, who will do the dirty work? Scott To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
Yes, moving to gcc32 is highly desirable for -current, otherwise we will be stuck at gcc311 for the entire life of FreeBSD 5.x. The important question to ask is, who will do the dirty work? Moving to GCC 3.2 will do us no good. The lifetime of the 3.2 release will be pretty short and 3.3 is supposed to replace it pretty soon. If we stick with 3.2 in -CURRENT, we'll find ourself tied to an old and unsupported release for the whole 5.x line, i.e. we'll risk to repeat 2.95.x story yet again. David O'Brien proposes to move -CURRENT directly to the 3.3 CVS shanshots, bypassing the GCC 3.2 version altogether. Early FreeBSD 5.x release(s) will not be polished for general consumption anyway, so that makes sense. By the time FreeBSD stabilizes, GCC 3.3 release will be ready. -- Alexander Kabaev To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
On Thu, Aug 15, 2002 at 11:59:11AM -0600, Long, Scott wrote: Hi, Are any plans to move to GCC 3.2 in current? Since it is just an ABI change it should work, without changing anything. It would give us a stable, multivendor ABI to work off of for the next line of 5.x releases. Just a thought. Jesse Gross Yes, moving to gcc32 is highly desirable for -current, otherwise we will be stuck at gcc311 for the entire life of FreeBSD 5.x. The important question to ask is, who will do the dirty work? Scott I'd be willing to help. I'm not exactly sure on what modifications to gcc are required to shove it into the base, but I have time (not working right now). :) -- -Erik [EMAIL PROTECTED] [http://math.smsu.edu/~erik] The opinions expressed by me are not necessarily opinions. In all probability, they are random rambling, and to be ignored. Failure to ignore may result in severe boredom or confusion. Shake well before opening. Keep Refrigerated. To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
RE: GCC 3.2
Yes, moving to gcc32 is highly desirable for -current, otherwise we will be stuck at gcc311 for the entire life of FreeBSD 5.x. The important question to ask is, who will do the dirty work? Moving to GCC 3.2 will do us no good. The lifetime of the 3.2 release will be pretty short and 3.3 is supposed to replace it pretty soon. If we stick with 3.2 in -CURRENT, we'll find ourself tied to an old and unsupported release for the whole 5.x line, i.e. we'll risk to repeat 2.95.x story yet again. David O'Brien proposes to move -CURRENT directly to the 3.3 CVS shanshots, bypassing the GCC 3.2 version altogether. Early FreeBSD 5.x release(s) will not be polished for general consumption anyway, so that makes sense. By the time FreeBSD stabilizes, GCC 3.3 release will be ready. I agree that gcc32 is not an ideal target either, but by going to it, we can upgrade to gcc33 when it's available and not loose binary compatibility (at least, according to the gcc folks). I'd rather move to gcc32 right now and get the binary compatibility pain out of the way, rather than wait for the last second to move to gcc33, then have to delay FreeBSD 5.0 because everything in c++ land is broken. Scott To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
I agree that gcc32 is not an ideal target either, but by going to it, we can upgrade to gcc33 when it's available and not loose binary compatibility (at least, according to the gcc folks). I'd rather move to gcc32 right now and get the binary compatibility pain out of the way, rather than wait for the last second to move to gcc33, then have to delay FreeBSD 5.0 because everything in c++ land is broken. The idea is to move to gcc 3.3-pre _now_ If GCC 3.2 has C++ ABI kinks worked out, GCC 3.3 surely has the same code in. GCC developers are trying to keep C++ ABI compatible between 3.2 and 3.3, but they are not giving any guaranrtees. -- Alexander Kabaev To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
Alexander Kabaev wrote: The idea is to move to gcc 3.3-pre _now_ If GCC 3.2 has C++ ABI kinks worked out, GCC 3.3 surely has the same code in. GCC developers are trying to keep C++ ABI compatible between 3.2 and 3.3, but they are not giving any guaranrtees. Cool. We can call it 3.3 in the release. Just like RedHat jumped the gun on the compiler release. And we all know how successful that was, right? -- Terry To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
Jesse Gross wrote: Are any plans to move to GCC 3.2 in current? Since it is just an ABI change it should work, without changing anything. It would give us a stable, multivendor ABI to work off of for the next line of 5.x releases. I believe David O'brien answer this the last 3 times it was asked. I will paraphrase: No, we are waiting for 3.3. -- Terry To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
Cool. We can call it 3.3 in the release. Terry, we will name it the same way we name our current GCC 3.1 snapshots. FreeBSD always shipped tweaked version of GCC with a bunch of local changes merges in. In STABLE, for example, we have gcc version 2.95.4 20020320 [FreeBSD] Just like RedHat jumped the gun on the compiler release. We are not _releasing_ our own version of GCC and we do not invent our own version numbers for it, so your attempt to compare us with RedHat is unjustified. Again, FreeBSD 5.0 will be in no shape for serious production use and putting GCC 3.2 there just to replace it with newer and possibly binary incompatible 3.3 release shortly afterwards is a complete waste of time. And we all know how successful that was, right? On the other side, we all know how successfull we were trying to get GCC 2.95.x bugs fixed for us, right? Do you really want to repeat this deeply satisfying experiment again? -- Alexander Kabaev To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
Alexander Kabaev wrote: We are not _releasing_ our own version of GCC and we do not invent our own version numbers for it, so your attempt to compare us with RedHat is unjustified. Again, FreeBSD 5.0 will be in no shape for serious production use and putting GCC 3.2 there just to replace it with newer and possibly binary incompatible 3.3 release shortly afterwards is a complete waste of time. There's always waiting for 3.3 to be released before trying to incorporate it... And we all know how successful that was, right? On the other side, we all know how successfull we were trying to get GCC 2.95.x bugs fixed for us, right? Do you really want to repeat this deeply satisfying experiment again? That was because the patches were not being submitted back against the unadulterated distribution code someone who had signed the assignment of rights to the FSF. The inability to get patches into 2.95 is totally unrelated to the fact that it was an older GCC, and completely related to the fact that the patches were not submitted in accordance with the GCC maintainer's guidelines, combined with not a little Linux advocacy and ELF advocacy. This issue is *nothing* like FreeBSD's steadfast refusal for *two years* to adopt ELF, and GCC treating non-ELF support as legacy support, with no expectations of continued developement. In the context the question was asked, it was *also* not about FreeBSD trying to get patches into GCC, it was about upgrading to GCC 3.2. It was also about trolling the mailing lists to cause just this sort of heated discussion (congradulations on playing into Jesse Gross's trolling here). It's all well and good to volunteer David O'Brien for additional *useless* work that he has already stated is *useless work*. I could understand raising the issue (though not over and over and over in a short period of time, as Mr. Gross has done recently) if the works was considered something that needed to be done immediately, or if patches to bmake the GCC 3.3 experimental release people want FreeBSD to user were being submitted, but all it's been so far is request for David O'Brien to do work he considers useless. FreeBSD has been conservative in its adoption of new compilers in the past; it would, in fact be reasonable, from an historical perspective, to not see 3.3 adopted for over a year following its release. I don't see why waiting for 3.3 to actually be released is such a terrible idea. Can *you* absolutely *guarantee* no binary incompatabilities between 3.3, as it sits now, in experimental form, and the final release of 3.3? If not, then I don't see why are exploding at me. -- Terry To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
That was because the patches were not being submitted back against the unadulterated distribution code someone who had signed the assignment of rights to the FSF. That was because GCC 2.95.x branch is closed for maintenance. The is no need in complex theory when a simple explanation is more than adequate. It was also about trolling the mailing lists to cause just this sort of heated discussion (congradulations on playing into Jesse Gross's trolling here). Sorry, guilty as charged. I was trying to get a people opinion on the issue. I will gladly volunteer to import a new version of GCC into -CURRENT myself, if there are no objections and if nobody is doing that already. I think David got a point though and I want his proposal to be discussed more. GCC 3.2 is an interim release and under no circumstances we should get tied to it through all the 5.x branch lifetime. No one will give a damn about it once 3.3 goes into maintenance. Can *you* absolutely *guarantee* no binary incompatabilities between 3.3, as it sits now, in experimental form, and the final release of 3.3? If not, then I don't see why are exploding at me. 3.1-pre to 3.2 upgrade breaks compatibility already. Can you guarantee that 3.3 will be backwards compatible with 3.2? This is yet another potential ABI breakage at the time when we'll be _forced_ to upgrade. How often do you expect GCC developers to break ABI with release scheduled to happed and the end of the year? -- Alexander Kabaev To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
It was also about trolling the mailing lists to cause just this sort of heated discussion (congradulations on playing into Jesse Gross's trolling here). This was *not* about trolling the mailing list. I wish I were intelligent enough to predict the behavior of thousands of people, most of whose names I don't even know, to cause a chain reaction to result in something like this. I do know that I did not intend for this particular result to happen, and am sorry I started this thread. Believe it or not, sometimes things are actually what they seem, in this case it really was a simple question. Terry, please do not bother replying to this message. Unless people wish to discuss technical details, this thread should be ended. Jesse Gross __ Do You Yahoo!? HotJobs - Search Thousands of New Jobs http://www.hotjobs.com To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: GCC 3.2
Alexander Kabaev wrote: Can *you* absolutely *guarantee* no binary incompatabilities between 3.3, as it sits now, in experimental form, and the final release of 3.3? If not, then I don't see why are exploding at me. 3.1-pre to 3.2 upgrade breaks compatibility already. Can you guarantee that 3.3 will be backwards compatible with 3.2? This is yet another potential ABI breakage at the time when we'll be _forced_ to upgrade. How often do you expect GCC developers to break ABI with release scheduled to happed and the end of the year? Once for every time the code is imported into FreeBSD, plus one. I think Murphy is a GCC committer... 8-). -- Terry To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message