Date: Fri, 14 Dec 2001 23:45:24 -0500 From: Mike Barcroft <mike@FreeBSD.ORG> To: Alexey Zelkin <phantom@FreeBSD.ORG> Cc: Bill Fenner <fenner@research.att.com>, standards@FreeBSD.ORG Subject: Re: cvs commit: src/lib/libc/stdlib strtod.c Message-ID: <20011214234524.J68310@espresso.q9media.com> In-Reply-To: <20011211111750.A35302@ark.cris.net>; from phantom@FreeBSD.ORG on Tue, Dec 11, 2001 at 11:17:50AM %2B0200 References: <200111301248.fAUCmV345233@freefall.freebsd.org> <20011130145450.B68788@ark.cris.net> <200112021729.JAA08989@windsor.research.att.com> <20011210115728.A11494@ark.cris.net> <20011210211240.M1956@espresso.q9media.com> <20011211111750.A35302@ark.cris.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Alexey Zelkin <phantom@FreeBSD.ORG> writes: > I don't see a reason to get it vendor imported for few reasons: > > 1. it's only one file (src/lib/libc/stdlib/strtod.c). Maybe we need > to think to take few more like strtof(), but I not sure in > full and generalized versions compatibility. Need testing. Most > of other stuff is not supposed to work on IEEE arifmetic machines > or not standard (not included to POSIX, SuS or C99). Correct me > if I wrong. I think we should try to utilize as much of this code as possible. > 2. we'll have to change it (at least for adding new definitions of > arches). Vendor's version is supposed to work on i8087, m86k, > vax, ibm and cray arches. Perhaps the vendor would be willing to integrate support for FreeBSD. > 3. And according of "much soon than us": All problems that Bruce > mentioned are not problems, just style improvements. From Message-ID <20011203062435.C9600-100000@gamplex.bde.org>: : Most of our local changes were in the vendor version 3-4 years before : we made them, sigh. Summary of some relevant parts of changes.gz from : the above URL: : - Jan 21 1993: apparently the long -> Long hack to support systems with : 32 bit ints and 64 bit longs : - Jul 6 1996: unsigned long -> ULong : - Dec 17 1996: threads support None of these issues appear to be style related; care to elaborate? > And last I think difference in slow reaction on addition of new > features/imporovement to Netlib (merging them to FreeBSD) is more > result of lack of interest to such complicated (and already working!) > code. So, let me some time and then you'll see "request for review", please > give comments. Right, that's why I'm suggesting a vendor maintained source, because no one has enough interest to maintain it in FreeBSD. > Even (1) is enough to understand that vendor import is just overhead > because file will be taken out of vendor branch as soon as powerpc and sparc64 > arches comes to "make world" as it already has been done for alpha and ia64. I would hardly be suggesting a vendor import if I thought we wouldn't be able to leave it on the vendor branch. Best regards, Mike Barcroft To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-standards" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20011214234524.J68310>