From mboxrd@z Thu Jan 1 00:00:00 1970 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me X-Spam-Level: X-Spam-Status: No, score=-1.3 required=5.0 tests=BAYES_00,INVALID_MSGID autolearn=no autolearn_force=no version=3.4.4 X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: 103376,1efdd369be089610 X-Google-Attributes: gid103376,public X-Google-Thread: 1025b4,499ea588f6acabff X-Google-Attributes: gid1025b4,public From: Ronald Cole Subject: Re: gnat-3.10 Date: 1997/07/02 Message-ID: #1/1 X-Deja-AN: 254269328 Sender: ronald@devo.ridgenet.net References: <1997Jun23.102715.1@eisner> Organization: RidgeNet - SLIP/PPP Internet, Ridgecrest, CA. (760) 371-3501 Newsgroups: comp.lang.ada,gnu.misc.discuss Date: 1997-07-02T00:00:00+00:00 List-Id: dewar@merv.cs.nyu.edu (Robert Dewar) writes: > As we all know, Ronald had trouble with the 3.09 > public release, perhaps that says we should have waited LONGER rathern > than SHORTER to distribute 3.09, though one cannot be sure that all > problems will be solved at any time. You put Richard Kenner's lastest patches in the publicly released gcc-272.dif the day before you released it. Clearly, they weren't well tested. So, if you aren't willing to build and test the public binary releases before you release the source, I would prefer that you didn't allow last-minute changes to your "well tested" prereleases as a matter of policy. Particularly, if you're not going to make "platform" fix releases (like gcc-2.7.2.1.tar.gz and emacs-19.34b.tar.gz were). I'll even go so far as to say that you could have just put a newer gcc-272.dif file on cs.nyu.edu with a note explaining the problem! Not quite a year ago, I was able to build a combined gcc-2.7.2, gnat-3.07, and g77-0.5.18 that bootstrapped perfectly with -O on HPUX. That is a quite a tribute to Richard Kenner and his work on gcc! I would hope to be able to do the same with gcc-2.7.2.2, gnat-3.10, and g77-0.5.20. I am willing to wait for a well-tested public 3.10, just like I'm willing to wait for gcc-2.8. I am very confident that when gcc-2.8.0 is finally released that it will bootstrap with -O2 on the HPUX platform! > Incidentally Richard Stallman is definitely supportive of the notion of > not releasing free software before it is in an appropriate state. The > Linux folks have often argued that the Linux distributes should contain > the latest GCC snapshots, since obviously they fix bugs, but such an > attitude is not a good idea in the long run. Linux is very stable because public "snapshots" are made available quite frequently for a lot of people to build and report problems they encounter. Linux development is truly a community project. I generally don't mess with the development releases until Linus declares that he is freezing features and readying the next stable version, at which point I lend a hand in making it very stable by building it on as many different platforms/configurations as I am able to. > other hand, Ronald's frustration with 3.09 is a good example of what we > want to avoid happening by making sure that public releases are in good > shape before they occur. I'd have to agree that it was pretty foolish to make a public release of 3.09 that depended on untested/unproven last-minute patches to gcc. I'll reserve judgement on 3.10 until you release it publicly. > This is simply a matter of quality control, which seems in our judgment > to be as important for public releases of GNAT as it is for any other > customer releases that we make. Then you might want to seriously consider building and testing the public binary releases before you release the source publicly. -- Forte International, P.O. Box 1412, Ridgecrest, CA 93556-1412 Ronald Cole Phone: (760) 499-9142 President, CEO Fax: (760) 499-9152 My PGP fingerprint: E9 A8 E3 68 61 88 EF 43 56 2B CE 3E E9 8F 3F 2B