Just to clarify, the suggestion to use #defines when --enable- gmpcompat is used, for functions we have deprecated and (eventually) removed in MPIR is obviously sensible (unless GMP also deprecates them, as we hope they will). My comment above about not including them with --enable-gmpcompat only applies to functions that GMP has deprecated (a long time ago).
We realise that by doing this we are effectively pushing upstream packages to change their code. We've already spoken to the authors of MPFR about this for example, and generally we found that people are happy to change code which uses long deprecated functions. Bill. On Jan 28, 11:55 pm, Bill Hart <goodwillh...@googlemail.com> wrote: > On Jan 28, 7:03 pm, Robert Bradshaw <rober...@math.washington.edu> > wrote: > > > > > > > On Jan 28, 2010, at 10:15 AM, William Stein wrote: > > > > On Thu, Jan 28, 2010 at 10:02 AM, Bill Hart <goodwillh...@googlemail.com > > > > wrote: > > >> Hi all, > > > >> it is with pleasure that we (finally) officially release MPIR 1.3.0. > > >> It is available at our websitehttp://www.mpir.org/ > > > >> Please note the following important things: > > > >> * I have been unable to get any tests to pass on ultrasparc2 > > >> machines, > > >> including t2 (solaris) and gcc54 (linux). I am confident there are > > >> not > > >> bugs in the ultrasparc assembly code (as we haven't changed it and it > > >> worked before). Thus if someone can tell me how to get MPIR to behave > > >> on these machines "out-of-the-box" I will make the necessary changes > > >> to MPIR. In the mean time, we have to officially consider MPIR broken > > >> on such machines (or the machines we test on to be broken). Note this > > >> is *not* a Solaris issue as MPIR builds just fine on Solaris. > > > > Hmmm. That doesn't sound good. In order not to annoy David Kirkby > > > off again, I guess this means we can't upgrade MPIR to version 1.3.0 > > > in Sage? We could at least make it an optional package. > > > I agree, though I hope the wait will be short. What has changed on t2 > > in the meantime? > > > >> * It is no longer necessary to issue make install gmp-compat, however > > >> it is still necessary to use --enable-gmpcompat with configure if you > > >> want a gmp.h and libgmp > > > >> * Many users believe that mpz_nextprime returns a prime. In fact it > > >> is > > >> only guaranteed to define a number which is with high probability > > >> prime. For this reason the function has been deprecated. Please use > > >> the new mpz_next_likely_prime in its place. > > > Isn't "probable prime" more standard terminology? > > No, that has a specific mathematical meaning. MPIR does not compute a > probable prime. > > > > > > > > > >> * Most of the random functions in MPIR which require global random > > >> state have been deprecated and new functions provided which have > > >> local > > >> state. This allows code using MPIR random functions to be threadsafe. > > > >> * The functions mpz_div_2exp and mpz_div_ui have been removed from > > >> MPIR. These were deprecated many years ago, but many people have > > >> continued to use them. > > > >> A quick workaround is to use the following defines: > > > >> #ifndef mpz_div_2exp > > >> #define mpz_div_2exp mpz_tdiv_q_2exp > > >> #endif > > > >> #ifndef mpz_div_ui > > >> #define mpz_div_ui mpz_tdiv_q_ui > > >> #endif > > > >> However, the better fix is to actually replace all occurrences in > > >> code. (Users of FLINT will have to wait a couple of days until I > > >> issue > > >> an update to FLINT, in particular the quadratic sieve, which still > > >> uses the old functions - in the mean time, just add the above defines > > >> to flint.h). > > > > Does the above mean that you have decided to just deprecate any > > > attempt at cross-compatibility with GMP? That's how my fuzzy "user > > > ears" see this. > > > Perhaps when configured with --enable-gmpcompat, it could include > > these (and any other similar) defines in gmp.h. > > That has been suggested, but we decided that it was an incorrect > approach. The functions have been deprecated in GMP too, for a very > long time. We are all just being slack if we keep using them in our > code. I will be fixing FLINT. > > Bill. -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
