Date: 2001-09-10 12:27:18
Greg Colvin <gcolvin_at_[hidden]> wrote on Mon Sep 10 10:49:58 2001:
| From: Paul A. Bristow <boost_at_[hidden]>
| > I believe that this is definitely the system to go for. It meets all
| > reasonable needs, except ...
| > those of us stuck with MSVC++ will have to manage without for now?
| No doubt some desperate and fiendishly clever Booster's will
| find a way to get this code "working" on MSVC++.
| > Should we use this as a 'gauntlet' to throw down to challenge Microsoft -
| > "make your compiler compliant enough to compile (and run!) this code!".
| > (or else take the C++ label off the product!)
| When Hell freezes over. I frequently see ignorant reporters
| decribe C++ as "the Microsoft language," as opposed to "the
| Sun language, Java."
| > > -----Original Message-----
| > > From: wb_at_[hidden] [mailto:wb_at_[hidden]]
| > > Sent: Wednesday, September 05, 2001 11:37 PM
| > > To: boost_at_[hidden]
| > > Subject: [boost] Quantity libraries: SIunits
| > >
| > > In anticipation of my return to the office, I have posted to
| > > http://home.fnal.gov/~wb/SItempl8.pdf
| > > a "pre-print" version of the paper that I will be presenting at the
| > > Template Programming Workshop at OOPSLA in October.
Thank you for your vote of confidence re SIunits.
Provided the maintainability of the code is preserved, I am always
happy to work with "fiendishly clever" people to accomplish reasonable
goals. Getting SIunits to work with more compilers is certainly, in my
opinion, a reasonable goal. I stand by my statement, however, that
incorporating significant workarounds for manifestly broken or outdated
compilers is an expensive burden. It also send the mistaken message
that programmers willingly contort otherwise-compliant code to
compensate for compilers' lack of standards compliance.
From the Boost Library Requirements and Guidelines web page:
"There is no requirement that a library run on C++ compilers which do
not conform to the ISO standard."
I strongly support this position. As the author of ISOcxx, Fermilab's
C++ portability package, I have done more than my share of providing
portability hacks, but enough is enough: I suffered a heart attack two
years ago because (in my opinion, with hindsight) of the
extraordinarily high stress levels that working with a certain
defective platform engendered.
In my view, our community of dedicated and committed programmers has
performed near miraculous feats and has demonstrated truly amazing
virtuosity in coping with crappy compilers. But think what else could
have been done with the same effort, what other libraries could have
been designed and developed, how much further along we all could have
been had our energies and creativity not been cruelly diverted from our
main mission and interests.
I understand that Boost seeks wide acceptance, and aims for a diversity
of platforms. Yes, I know we all have to make a living. But I've been
programming for almost forty years, and I've now paid a price that I
want no one else to have to pay. Let's not get that "desperate"!
If SIunits turns out to be useful as a "gauntlet" for all compiler
vendors, I will feel my efforts will have been repaid. However, I
would ask that we focus, for now, on issues of more immediate concern
to SIunits' design and implementation.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk