[issue12450] Use the Grisu algorithms to convert floats to strings

Changes by STINNER Victor : -- type: enhancement -> performance ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe:

[issue12450] Use the Grisu algorithms to convert floats to strings

Bart Robinson added the comment: Six years later, I have accepted Amaury's challenge and created an extension module that uses the double-conversion library to generate repr's for floats: https://pypi.python.org/pypi/frepr My rudimentary performance testing gives something like 8X speedup comp

[issue12450] Use the Grisu algorithms to convert floats to strings

Mark Dickinson added the comment: > It's biggest deficiency (compared to Gay's dtoa.c) is its specialization > to > IEEE doubles. We're only using the portion of Gay's code (with some significant modifications at this point) that applies to the IEEE 754 binary64 format, so I don't think this

[issue12450] Use the Grisu algorithms to convert floats to strings

Florian Loitsch added the comment: FYI: the double-conversion library at http://code.google.com/p/double-conversion already contains code for the fallback case. It would not be necessary to keep Python's existing code just for the 0.5%. The library is now used by Firefox, Chrome, and Webkit,

[issue12450] Use the Grisu algorithms to convert floats to strings

Amaury Forgeot d'Arc added the comment: Agreed. If some volunteer wants to work on it, I suggest to make an extension module first, so that everybody can try and compare with the current routines. -- ___ Python tracker

[issue12450] Use the Grisu algorithms to convert floats to strings

Raymond Hettinger added the comment: I concur with Mark. This is a very challenging area and no change would be warranted without keeping the current accuracy guarantees, definitive speed improvement, extensive tests, and consideration of the impact on users of having float reprs change yet

[issue12450] Use the Grisu algorithms to convert floats to strings

Mark Dickinson added the comment: [Amaury] "It would be interesting to see if it is better/faster than the current dtoa." I agree that it would be interesting to compare. [Eric] "maybe we can say we we can live with 99.5% shortest repr coverage" Please no! As you say, we'd still need the fal

[issue12450] Use the Grisu algorithms to convert floats to strings

Changes by STINNER Victor : -- nosy: +haypo ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.python.

[issue12450] Use the Grisu algorithms to convert floats to strings

Raymond Hettinger added the comment: FWIW, I don't think this is worth further disruption. We already have an excellent set of float repr routines. There is no problem to be solved here. -- assignee: -> rhettinger priority: normal -> low versions: +Python 3.3 __

[issue12450] Use the Grisu algorithms to convert floats to strings

Eric V. Smith added the comment: I see the problems as: 1. Given Python's other overhead, we'd need to profile to show an improvement in the speed of this conversion would make a noticeable impact on any import workload. 2. If we want to keep the shortest-float-repr property for all possible

[issue12450] Use the Grisu algorithms to convert floats to strings

Changes by Eric V. Smith : -- nosy: +eric.smith ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.pyt

[issue12450] Use the Grisu algorithms to convert floats to strings

Changes by Dirkjan Ochtman : -- nosy: +djc ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.python.o

[issue12450] Use the Grisu algorithms to convert floats to strings

New submission from Amaury Forgeot d'Arc : Reported by Michael Ford in msg139402: http://www.serpentine.com/blog/2011/06/29/here-be-dragons-advances-in-problems-you-didnt-even-know-you-had/ describes a new algorithm for float<->str conversions. It would be interesting to see if it is better/fast