Re: [Python-3000] __format__ and datetime

2007-09-12 Thread Skip Montanaro
> Given how long strftime has been around I think %f is fine. We may > even influence the future of the C library. :-) Patch for datetime (py3k only at this point, no tests either) here: http://bugs.python.org/issue1158 Skip ___ Python-3000 mailing

Re: [Python-3000] __format__ and datetime

2007-09-12 Thread skip
Guido> Which ones are two letters? All the locale-specific stuff on Solaris 10. I guess technically the first letter of the pair is a modifier of the actual code, which comes next. From the man page: Modified Conversion Specifications Some conversion specifications can be modified b

Re: [Python-3000] Solaris support in 3.0?

2007-09-12 Thread Joe Smith
"Paul Moore" <[EMAIL PROTECTED]> wrote in message news:[EMAIL PROTECTED] > On 05/09/07, Gregory P. Smith <[EMAIL PROTECTED]> wrote: >> Rather than resurrecting the old RSA-copyright md5.c I can easily make >> new >> ones out of the libtomcrypt md5 and sha1 sources the same way i created >> the

Re: [Python-3000] Performance Notes - new hash algorithm

2007-09-12 Thread Nicko van Someren
On 10 Sep 2007, at 01:58, Jim Jewett wrote: > To spell this out a bit more: > ... > When adding four entries to an 8-slot table, a truly random hash would > have at least one collision (0/8 + 1/8 + 2/8 + 3/8 =) 3/4 of the > time. As expected, the proposed hash does have a collision for those > fo

Re: [Python-3000] patch: bytes object PyBUF_LOCKDATA read-only and immutable support

2007-09-12 Thread Guido van Rossum
That's a different topic altogether. We're talking here about locking modes for the buffer API (PEP 3118). This does not involve actual locks -- the operations just fail if the requested lock cannot be obtained. On 9/12/07, Joel Bender <[EMAIL PROTECTED]> wrote: > > Sounds much like the modes offe

Re: [Python-3000] C API for ints and strings

2007-09-12 Thread Guido van Rossum
On 9/12/07, Nicholas Bastin <[EMAIL PROTECTED]> wrote: > On 9/12/07, Guido van Rossum <[EMAIL PROTECTED]> wrote: > > Can I just shortcut this discussion saying that we will *not* switch > > to use GMP? It's just not going to happen. Period. End of discussion. > > I figured that was assumed once it

Re: [Python-3000] C API for ints and strings

2007-09-12 Thread Nicholas Bastin
On 9/12/07, Guido van Rossum <[EMAIL PROTECTED]> wrote: > Can I just shortcut this discussion saying that we will *not* switch > to use GMP? It's just not going to happen. Period. End of discussion. I figured that was assumed once it was pointed out that it didn't work on Intel macs... I'm pretty

Re: [Python-3000] patch: bytes object PyBUF_LOCKDATA read-only and immutable support

2007-09-12 Thread Joel Bender
> Sounds much like the modes offered by an old operating system that had a > very nice lock manager. Awe, VMS isn't THAT old, is it? :-) I have a wrapper around threading.Lock and threading.RLock that I've been using that does deadlock detection and have wished for these lock modes many times

Re: [Python-3000] patch: bytes object PyBUF_LOCKDATA read-only and immutable support

2007-09-12 Thread Russell E. Owen
In article <[EMAIL PROTECTED]>, "Guido van Rossum" <[EMAIL PROTECTED]> wrote: > I guess I would be inclined to propose separate flags for indicating > the operation that the caller will attempt (read or write) and the > level of locking (lock the buffer's address or also prevent anyone > else fr

Re: [Python-3000] __format__ and datetime

2007-09-12 Thread Guido van Rossum
On 9/12/07, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote: > So, is '%f" okay to coopt? Is there some sort of future-proofing we can do > so that if the libc folks decide later to use "%f" for something we're not > (mildly) hosed? Maybe "%."? It appears that all strftime codes are one or > two let

Re: [Python-3000] C API for ints and strings

2007-09-12 Thread Guido van Rossum
Can I just shortcut this discussion saying that we will *not* switch to use GMP? It's just not going to happen. Period. End of discussion. -- --Guido van Rossum (home page: http://www.python.org/~guido/) ___ Python-3000 mailing list Python-3000@python.o

Re: [Python-3000] C API for ints and strings

2007-09-12 Thread Barry Warsaw
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On Sep 12, 2007, at 1:43 PM, Neil Schemenauer wrote: > Larry Hastings <[EMAIL PROTECTED]> wrote: >> I am opposed to using LGPL- or GPL-licensed code in Python. > > Me too. Also, I don't see the point. Python's current long integer > performance is g

Re: [Python-3000] C API for ints and strings

2007-09-12 Thread Neil Schemenauer
Larry Hastings <[EMAIL PROTECTED]> wrote: > I am opposed to using LGPL- or GPL-licensed code in Python. Me too. Also, I don't see the point. Python's current long integer performance is good enough for the large majority of Python users. For the few specialized users, an extension module should

Re: [Python-3000] __format__ and datetime

2007-09-12 Thread skip
Guido> No, the datetime module is explicitly defined to use Guido> microseconds. I don't expect there to be a practical use for Guido> nanoseconds (even microseconds are doubtful, but useful since one Guido> might want unique timestamps for more than 1000 events per Guido> seco

Re: [Python-3000] patch: bytes object PyBUF_LOCKDATA read-only and immutable support

2007-09-12 Thread Gregory P. Smith
On 9/11/07, Greg Ewing <[EMAIL PROTECTED]> wrote: > > Guido van Rossum wrote: > > Any number of concurrent SIMPLE accesses can > > coexist since the clients promise they will only read. > > As a general principle, using a word like SIMPLE in an > API is a really bad idea imo, as it's far too vague.