Bug #34584 for Math-BigRat: hang when calling exp

Mon Mar 31 16:38:27 2008 steve [...] mouf.net - Ticket created

Subject:	hang when calling exp
Date:	Mon, 31 Mar 2008 16:37:48 -0400
To:	bug-Math-BigInt [...] rt.cpan.org
From:	Steve Wills <steve [...] mouf.net>

With just perl 5.8.8 installed, I can run this without problems: #!/usr/local/bin/perl use strict; use Math::BigRat; my $rat = Math::BigRat->new( '1 / 2' ); print "$rat\n"; my $exp = exp $rat; print "$exp\n"; exit 0; However, if I install Math::BigInt 1.87, this hangs at the call to exp, with perl using 100% CPU. I'm also told this fails with perl 5.10 and Math::BigInt 1.88, but have not tested personally. Thanks, Steve

Tue Apr 01 12:41:35 2008 TELS [...] cpan.org - Taken

Sat Apr 05 17:42:58 2008 TELS [...] cpan.org - Correspondence added

From:

TELS [...] cpan.org

Hi Steve, (via RT so the mail gets recored) So far I found out that it hangs in this operation: $x->bpow($x_org, @params) where $x is: 90933395208605785401971970164779391644753259799242 / 33452526613163807108170062053440751665152000000000 and $x is 1/2. This means it needs to take the square root of $x. After a bit of debugging it turns out the general algorithm can't deal with "A/B ** 1/N" - I now added a shortcut that routes these cases through bsqrt() (for 1/2) or broot (for 1/N where N > 2). I made a pre-release package, can you please test it? http://Bloodgate.com/perl/packages/devel/Math-BigRat-0.22.tar.gz Thanx! Tels

Sat Apr 05 17:43:03 2008 The RT System itself - Status changed from 'new' to 'open'

Sat Apr 05 17:47:54 2008 TELS [...] cpan.org - TimeWorked changed from (no value) to '25'

Sat Apr 05 17:47:58 2008 TELS [...] cpan.org - Queue changed from Math-BigInt to Math-BigRat

Sat Apr 05 17:48:40 2008 TELS [...] cpan.org - Broken in 0.20 added

Sat Apr 05 17:48:40 2008 TELS [...] cpan.org - Severity Important added

Sun Apr 06 00:51:29 2008 steve [...] mouf.net - Correspondence added

Subject:	Re: [rt.cpan.org #34584] hang when calling exp
Date:	Sun, 6 Apr 2008 00:50:30 -0400
To:	bug-Math-BigInt [...] rt.cpan.org
From:	Steve Wills <steve [...] mouf.net>

HI there, On Apr 5, 2008, at 5:43 PM, TELS via RT wrote: Show quoted text

> > <URL: http://rt.cpan.org/Ticket/Display.html?id=34584 > > > Hi Steve, > > (via RT so the mail gets recored) > > So far I found out that it hangs in this operation: > > $x->bpow($x_org, @params) > > where $x is: > > 90933395208605785401971970164779391644753259799242 / > 33452526613163807108170062053440751665152000000000 > > and $x is 1/2. This means it needs to take the square root of $x. > > After a bit of debugging it turns out the general algorithm can't deal > with "A/B ** 1/N" - I now added a shortcut that routes these cases > through bsqrt() (for 1/2) or broot (for 1/N where N > 2). > > I made a pre-release package, can you please test it? > > http://Bloodgate.com/perl/packages/devel/Math- > BigRat-0.22.tar.gz

Thanks for taking a look. I upgraded my Math::BigInt, then ran the test code from the bug report. It now doesn't hang, but the results aren't what I expect: # perl exp.pl 1/2 9535900335500879457687887524133067574481 /5783815921445270815783609372070483523265 On a system with the older perl, I get: # perl exp.pl 1/2 1.64872127070013 (here, perl is 5.8.6 and Math::BigRat is 0.13) I'm no math expert, but the answer the folks I'm working with I believe expect a decimal answer as the older version did. Also, and again, I'm no math expert, what about broot cases where 1/N and N < 2? If I change the test case to 1/1.5, it still hangs. Again, thanks for taking a look at this, I greatly appreciate it. Steve

Sun Apr 06 05:23:34 2008 nospam-abuse [...] bloodgate.com - Correspondence added

Subject:	Re: [rt.cpan.org #34584] hang when calling exp
Date:	Sun, 6 Apr 2008 11:08:23 +0200
To:	bug-Math-BigRat [...] rt.cpan.org
From:	Tels <nospam-abuse [...] bloodgate.com>

Moin Steve, On Sunday 06 April 2008 06:51:41 Steve Wills via RT wrote: Show quoted text

> Queue: Math-BigRat > Ticket <URL: http://rt.cpan.org/Ticket/Display.html?id=34584 > > > HI there, > > On Apr 5, 2008, at 5:43 PM, TELS via RT wrote:

> > <URL: http://rt.cpan.org/Ticket/Display.html?id=34584 > > > > > Hi Steve, > > > > (via RT so the mail gets recored) > > > > So far I found out that it hangs in this operation: > > > > $x->bpow($x_org, @params) > > > > where $x is: > > > > 90933395208605785401971970164779391644753259799242 / > > 33452526613163807108170062053440751665152000000000 > > > > and $x is 1/2. This means it needs to take the square root of $x. > > > > After a bit of debugging it turns out the general algorithm can't > > deal with "A/B ** 1/N" - I now added a shortcut that routes these > > cases through bsqrt() (for 1/2) or broot (for 1/N where N > 2). > > > > I made a pre-release package, can you please test it? > > > > http://Bloodgate.com/perl/packages/devel/Math- > > BigRat-0.22.tar.gz

> > Thanks for taking a look. I upgraded my Math::BigInt, then ran the

You mean you upgraded your Math::BigRat? ;) Show quoted text

> test code from the bug report. It now doesn't hang, but the results > aren't what I expect: > > # perl exp.pl > 1/2 > 9535900335500879457687887524133067574481 > /5783815921445270815783609372070483523265

Well: # perl -Mbignum -wle 'print 9535900335500879457687887524133067574481 / 5783815921445270815783609372070483523265' 1.648721270700128146848650787814163571654 :-D Math::BigRat routines will generally return Math::BigRat objects - if they don't, that's a bug (or missing feature :) You should be able to do normal math with the returned object - that is the point of Math::BigRat after all. You can, of course, convert this number to a scalar, but you will lose digits: # perl -Mbigrat -wle 'my $a = 9535900335500879457687887524133067574481 / 5783815921445270815783609372070483523265; print $a->numify()' 1.64872127070013 Show quoted text

> On a system with the older perl, I get: > > # perl exp.pl > 1/2 > 1.64872127070013 > > (here, perl is 5.8.6 and Math::BigRat is 0.13)

That's a bug then :) Fromt the changelog: 2007-06-30 v0.20 Tels 956 tests * require Math::BigInt v1.87 * add bnok() and bexp() Hence bexp() and overloading for exp() was added in v0.20 of BigRat. If you want to use the "normal" exp() from Perl, it is available as CORE::exp(); perl -Mbigrat -wle 'print CORE::exp("0.5")' 1.64872127070013 Note the quoting of the argument to prevent "bigrat" from turning it into a Math::BigRat. (This is not nec. if you use Math::BigRat alone, tho) Show quoted text

> I'm no math expert, but the answer the folks I'm working with I > believe expect a decimal answer as the older version did. > > Also, and again, I'm no math expert, what about broot cases where 1/N > and N < 2? If I change the test case to 1/1.5, it still hangs.

Ah. Good catch. I think I need to refine the algorithm to handle these cases, too. Show quoted text

> Again, thanks for taking a look at this, I greatly appreciate it.

No, thank you for your report! All the best, Tels -- Signed on Sun Apr 6 11:01:12 2008 with key 0x93B84C15. Get one of my photo posters: http://bloodgate.com/posters PGP key on http://bloodgate.com/tels.asc or per email. "My glasses, my glasses. I cannot see without my glasses." - "My glasses, my glasses. I cannot be seen without my glasses."

Download (untitled)
application/pgp-signature 481b

Message body not shown because it is not plain text.

Sun Apr 06 08:20:03 2008 nospam-abuse [...] bloodgate.com - Correspondence added

Subject:	Re: [rt.cpan.org #34584] hang when calling exp
Date:	Sun, 6 Apr 2008 14:18:46 +0200
To:	bug-Math-BigRat [...] rt.cpan.org
From:	Tels <nospam-abuse [...] bloodgate.com>

Moin, On Sunday 06 April 2008 11:23:43 nospam-abuse@bloodgate.com via RT wrote: Show quoted text

> Queue: Math-BigRat > Ticket <URL: http://rt.cpan.org/Ticket/Display.html?id=34584 > > > Moin Steve,

> > Also, and again, I'm no math expert, what about broot cases where > > 1/N and N < 2? If I change the test case to 1/1.5, it still hangs.

> Ah. Good catch. I think I need to refine the algorithm to handle > these cases, too.

I updated the code at: http://bloodgate.com/perl/packages/devel/Math-BigRat-0.22.tar.gz It contains now an improved bpow() method, which handles cases like "1/5" as well as "2/3". I also fixed a bug in broot(), and I added a as_float() method which you can use to convert a Math::BigRat object into a Math::BigFloat object with a specified accuracy (e.g. to "X" digits). Hope this solves your problem now :) All the best, Tels -- Signed on Sun Apr 6 14:16:07 2008 with key 0x93B84C15. View my photo gallery: http://bloodgate.com/photos PGP key on http://bloodgate.com/tels.asc or per email. "Those are my principles, and if you don't like them... well, I have others." -- Groucho Marx

Download (untitled)
application/pgp-signature 481b

Message body not shown because it is not plain text.

Sun Apr 06 11:35:56 2008 steve [...] mouf.net - Correspondence added

Subject:	Re: [rt.cpan.org #34584] hang when calling exp
Date:	Sun, 6 Apr 2008 11:34:49 -0400
To:	bug-Math-BigRat [...] rt.cpan.org
From:	Steve Wills <steve [...] mouf.net>

On Apr 6, 2008, at 5:23 AM, nospam-abuse@bloodgate.com via RT wrote: Show quoted text

> > <URL: http://rt.cpan.org/Ticket/Display.html?id=34584 > > > Moin Steve, >

>> >> Thanks for taking a look. I upgraded my Math::BigInt, then ran the

> > You mean you upgraded your Math::BigRat? ;) >

Actually, I meant to say I upgraded both, since the updated Math::BigRat requires an updated Math::BigInt. Show quoted text

> Well: > > # perl -Mbignum -wle 'print > 9535900335500879457687887524133067574481 / > 5783815921445270815783609372070483523265' > 1.648721270700128146848650787814163571654 > > :-D >

Right, but the older version used to return the decimal. Show quoted text

> Math::BigRat routines will generally return Math::BigRat objects - if > they don't, that's a bug (or missing feature :) > > You should be able to do normal math with the returned object - that > is > the point of Math::BigRat after all. > > You can, of course, convert this number to a scalar, but you will lose > digits: > > # perl -Mbigrat -wle 'my $a = > 9535900335500879457687887524133067574481 / > 5783815921445270815783609372070483523265; print $a->numify()' > > 1.64872127070013 >

>> On a system with the older perl, I get: >> >> # perl exp.pl >> 1/2 >> 1.64872127070013 >> >> (here, perl is 5.8.6 and Math::BigRat is 0.13)

> > That's a bug then :) >

Ok. Show quoted text

> Fromt the changelog: > > 2007-06-30 v0.20 Tels 956 tests > * require Math::BigInt v1.87 > * add bnok() and bexp() > > Hence bexp() and overloading for exp() was added in v0.20 of BigRat. > > If you want to use the "normal" exp() from Perl, it is available as > CORE::exp(); > > perl -Mbigrat -wle 'print CORE::exp("0.5")' > 1.64872127070013 > > Note the quoting of the argument to prevent "bigrat" from turning it > into a Math::BigRat. (This is not nec. if you use Math::BigRat alone, > tho) >

Ok, noted. Perhaps we rely on a bug and need to update our code. Show quoted text

>> I'm no math expert, but the answer the folks I'm working with I >> believe expect a decimal answer as the older version did. >> >> Also, and again, I'm no math expert, what about broot cases where 1/N >> and N < 2? If I change the test case to 1/1.5, it still hangs.

> > Ah. Good catch. I think I need to refine the algorithm to handle these > cases, too.

Ok. Thanks, Steve

Sun Apr 06 12:07:17 2008 steve [...] mouf.net - Correspondence added

Subject:	Re: [rt.cpan.org #34584] hang when calling exp
Date:	Sun, 6 Apr 2008 12:06:44 -0400
To:	bug-Math-BigRat [...] rt.cpan.org
From:	Steve Wills <steve [...] mouf.net>

On Apr 6, 2008, at 8:20 AM, nospam-abuse@bloodgate.com via RT wrote: Show quoted text

>

>> Moin Steve,

>>> Also, and again, I'm no math expert, what about broot cases where >>> 1/N and N < 2? If I change the test case to 1/1.5, it still hangs.

>> Ah. Good catch. I think I need to refine the algorithm to handle >> these cases, too.

> > I updated the code at: > > http://bloodgate.com/perl/packages/devel/Math-BigRat-0.22.tar.gz > > It contains now an improved bpow() method, which handles cases > like "1/5" as well as "2/3". I also fixed a bug in broot(), and I > added > a as_float() method which you can use to convert a Math::BigRat object > into a Math::BigFloat object with a specified accuracy (e.g. to "X" > digits).

Yes it does, thanks. One thing I notice is that with the older perl and this code: my $rat = Math::BigRat->new( '2 / 3' ); my $exp = exp $rat; print $exp , "\n"; produces: 1.94773404105468 But with the newer Math::BigRat and this code (which I believe uses the new code correctly): my $rat = Math::BigRat->new( '2 / 3' ); my $exp = exp $rat; print $exp->as_float(15),"\n"; produces: 4.48168907033806 I don't know which is correct. Changing it to 1/2 or 1/3 produces the same results on both versions. Steve

Sun Apr 06 12:55:49 2008 nospam-abuse [...] bloodgate.com - Correspondence added

Subject:	Re: [rt.cpan.org #34584] hang when calling exp
Date:	Sun, 6 Apr 2008 18:54:39 +0200
To:	bug-Math-BigRat [...] rt.cpan.org
From:	Tels <nospam-abuse [...] bloodgate.com>

Moin, On Sunday 06 April 2008 18:07:27 Steve Wills via RT wrote: [snip] Show quoted text

> >> Ah. Good catch. I think I need to refine the algorithm to handle > >> these cases, too.

> > > > I updated the code at: > > > > http://bloodgate.com/perl/packages/devel/Math-BigRat-0.22.tar.gz > > > > It contains now an improved bpow() method, which handles cases > > like "1/5" as well as "2/3". I also fixed a bug in broot(), and I > > added > > a as_float() method which you can use to convert a Math::BigRat > > object into a Math::BigFloat object with a specified accuracy (e.g. > > to "X" digits).

> > Yes it does, thanks. > > One thing I notice is that with the older perl and this code: > > my $rat = Math::BigRat->new( '2 / 3' ); > my $exp = exp $rat; > print $exp , "\n"; > > produces: > > 1.94773404105468 > > But with the newer Math::BigRat and this code (which I believe uses > the new code correctly): > > my $rat = Math::BigRat->new( '2 / 3' ); > my $exp = exp $rat; > print $exp->as_float(15),"\n"; > > produces: > > 4.48168907033806 > > I don't know which is correct. Changing it to 1/2 or 1/3 produces the > same results on both versions.

My pocket calculator says 1.9477... is correct, so the new code is still buggy. Sorry! I have another look, Tels PS: If your old code relied on exp() returning a scalar, then yes, the code needs to be changed. I am not sure why you need a scalar, but the Math::BigRat module aims (like the other ones) to be transparent. Although it is of course slower than normal scalars. Also, Math::BigRat is not yet fully developed like the other Math::Big* modules. :) -- Signed on Sun Apr 6 18:52:43 2008 with key 0x93B84C15. Get one of my photo posters: http://bloodgate.com/posters PGP key on http://bloodgate.com/tels.asc or per email. "Blogebrity: Wow, guess what this one stands for? Too easy. Hey, anyone can do it: take a blogger who's a chef, and you get: BLEF. A blogger who's a dentist? BENTIST. A female blogger with an itch? You guessed it: a BITCH." -- maddox from xmission

Download (untitled)
application/pgp-signature 481b

Message body not shown because it is not plain text.

Sun Apr 06 13:19:40 2008 nospam-abuse [...] bloodgate.com - Correspondence added

Subject:	Re: [rt.cpan.org #34584] hang when calling exp
Date:	Sun, 6 Apr 2008 19:17:49 +0200
To:	bug-Math-BigRat [...] rt.cpan.org
From:	Tels <nospam-abuse [...] bloodgate.com>

Moin Steve, On Sunday 06 April 2008 18:07:27 Steve Wills via RT wrote: Show quoted text

> Queue: Math-BigRat > Ticket <URL: http://rt.cpan.org/Ticket/Display.html?id=34584 > > > One thing I notice is that with the older perl and this code: > > my $rat = Math::BigRat->new( '2 / 3' ); > my $exp = exp $rat; > print $exp , "\n"; > > produces: > > 1.94773404105468 > > But with the newer Math::BigRat and this code (which I believe uses > the new code correctly): > > my $rat = Math::BigRat->new( '2 / 3' ); > my $exp = exp $rat; > print $exp->as_float(15),"\n"; > > produces: > > 4.48168907033806

I fixed that in the newest revision (I swapped by accident denominator and nominator - oops). However, since the operation of $x->broot($y) is basically: A/B ** C/D => broot( A ** D / B ** D, C) it turns out that if both A and D, or B and D are very big, the resulting temp. number gets huge. For instance: The result is always the same (despite looking a bit different in every case), but the time rises fast: # time perl -Ilib -MMath::BigRat -wle 'print Math::BigRat->new(2.71828182845905) ** Math::BigRat->new("67/100")' 390847464127188358237183659254469213981/200000000000000000000000000000000000000 real 0m0.229s user 0m0.216s sys 0m0.012s # time perl -Ilib -MMath::BigRat -wle 'print Math::BigRat->new(2.71828182845905) ** Math::BigRat->new("667/1000")' 1948383393954500470047617243577940984231/1000000000000000000000000000000000000000 real 0m0.463s user 0m0.456s sys 0m0.008s # time perl -Ilib -MMath::BigRat -wle 'print Math::BigRat->new(2.71828182845905) ** Math::BigRat->new("6667/10000")' 77911958664191991164081333000359819941/40000000000000000000000000000000000000 real 0m29.331s user 0m29.030s sys 0m0.040s Clearly not optimal... All the best, Tels -- Signed on Sun Apr 6 19:06:38 2008 with key 0x93B84C15. View my photo gallery: http://bloodgate.com/photos PGP key on http://bloodgate.com/tels.asc or per email. "HOT PACKET ON SERVER ACTION! Click here for FREE ACCESS to streaming video of dirty packets penetrating badly-configured firewalls!!!" -- aanand (705284) on 2004-03-20 on /. about the Adult and Evil bits

Download (untitled)
application/pgp-signature 481b

Message body not shown because it is not plain text.

Sun Apr 06 15:11:11 2008 steve [...] mouf.net - Correspondence added

Subject:	Re: [rt.cpan.org #34584] hang when calling exp
Date:	Sun, 6 Apr 2008 15:09:21 -0400
To:	bug-Math-BigRat [...] rt.cpan.org
From:	Steve Wills <steve [...] mouf.net>

On Apr 6, 2008, at 1:20 PM, nospam-abuse@bloodgate.com via RT wrote: Show quoted text

> > > I fixed that in the newest revision (I swapped by accident denominator > and nominator - oops).

Great! Thanks. Btw, just wondering, are you adding test cases for these? Show quoted text

> However, since the operation of $x->broot($y) is > basically: > > A/B ** C/D => broot( A ** D / B ** D, C) > > it turns out that if both A and D, or B and D are very big, the > resulting temp. number gets huge. For instance: > > The result is always the same (despite looking a bit different in > every > case), but the time rises fast: > > # time perl -Ilib -MMath::BigRat -wle 'print > Math::BigRat->new(2.71828182845905) ** Math::BigRat->new("67/100")' > 390847464127188358237183659254469213981 > /200000000000000000000000000000000000000 > > real 0m0.229s > user 0m0.216s > sys 0m0.012s > # time perl -Ilib -MMath::BigRat -wle 'print > Math::BigRat->new(2.71828182845905) ** Math::BigRat->new("667/1000")' > 1948383393954500470047617243577940984231 > /1000000000000000000000000000000000000000 > > real 0m0.463s > user 0m0.456s > sys 0m0.008s > # time perl -Ilib -MMath::BigRat -wle 'print > Math::BigRat->new(2.71828182845905) ** Math::BigRat-

> >new("6667/10000")'

> 77911958664191991164081333000359819941 > /40000000000000000000000000000000000000 > > real 0m29.331s > user 0m29.030s > sys 0m0.040s > > Clearly not optimal...

Yes. With my test code: #!/usr/local/bin/perl use strict; use Math::BigRat; my $rat = Math::BigRat->new( '67 / 100' ); my $exp = exp $rat; print $exp->as_float(15) , "\n"; # time perl exp.pl 1.95423732063594 1.239u 0.029s 0:01.26 99.2% 10+1165k 0+0io 0pf+0w With 667/1000: # time perl exp.pl 1.94838339395450 79.383u 0.037s 1:19.45 99.9% 10+1137k 0+0io 0pf+0w With 6667/10000, well, I got tired of waiting on it after 30 minutes. Is there anything that can be done to improve the performance? Steve

Sun Apr 06 16:27:08 2008 nospam-abuse [...] bloodgate.com - Correspondence added

Subject:	Re: [rt.cpan.org #34584] hang when calling exp
Date:	Sun, 6 Apr 2008 22:26:19 +0200
To:	bug-Math-BigRat [...] rt.cpan.org
From:	Tels <nospam-abuse [...] bloodgate.com>

On Sunday 06 April 2008 21:11:33 Steve Wills via RT wrote: Show quoted text

> Queue: Math-BigRat > Ticket <URL: http://rt.cpan.org/Ticket/Display.html?id=34584 > > > On Apr 6, 2008, at 1:20 PM, nospam-abuse@bloodgate.com via RT wrote:

> > I fixed that in the newest revision (I swapped by accident > > denominator and nominator - oops).

> > Great! Thanks. Btw, just wondering, are you adding test cases for > these?

Yeah, this time I do :) (I added tests, but not for this specific case before) [snip] Show quoted text

> > Clearly not optimal...

> > Yes. With my test code: > > #!/usr/local/bin/perl > > use strict; > use Math::BigRat; > > my $rat = Math::BigRat->new( '67 / 100' ); > my $exp = exp $rat; > print $exp->as_float(15) , "\n"; > > # time perl exp.pl > 1.95423732063594 > 1.239u 0.029s 0:01.26 99.2% 10+1165k 0+0io 0pf+0w > > With 667/1000: > > # time perl exp.pl > 1.94838339395450 > 79.383u 0.037s 1:19.45 99.9% 10+1137k 0+0io 0pf+0w > > With 6667/10000, well, I got tired of waiting on it after 30 minutes. > > Is there anything that can be done to improve the performance?

Yeah I think so, but I need firt to think of an algorithm to calculate the bpow() result without losing precision. Probably will release the next version just as it is, because I don't know how long it will take to do this. Thanx for the report again, Tels -- Signed on Sun Apr 6 21:27:57 2008 with key 0x93B84C15. Get one of my photo posters: http://bloodgate.com/posters PGP key on http://bloodgate.com/tels.asc or per email. "I have always wished that my computer would be as easy to use as my telephone. My wish has come true. I no longer know how to use my telephone." -- Bjarne Stroustrup, inventor of the C++ programming language

Download (untitled)
application/pgp-signature 481b

Message body not shown because it is not plain text.

Mon Apr 07 14:42:28 2008 steve [...] mouf.net - Correspondence added

Subject:	Re: [rt.cpan.org #34584] hang when calling exp
Date:	Mon, 7 Apr 2008 14:41:56 -0400
To:	bug-Math-BigRat [...] rt.cpan.org
From:	Steve Wills <steve [...] mouf.net>

Hi there, Just wondering, when are you planning to release 0.22 to CPAN? Thanks, Steve On Apr 6, 2008, at 4:27 PM, nospam-abuse@bloodgate.com via RT wrote: Show quoted text

> > <URL: http://rt.cpan.org/Ticket/Display.html?id=34584 > > > On Sunday 06 April 2008 21:11:33 Steve Wills via RT wrote:

>> Queue: Math-BigRat >> Ticket <URL: http://rt.cpan.org/Ticket/Display.html?id=34584 > >> >> On Apr 6, 2008, at 1:20 PM, nospam-abuse@bloodgate.com via RT wrote:

>>> I fixed that in the newest revision (I swapped by accident >>> denominator and nominator - oops).

>> >> Great! Thanks. Btw, just wondering, are you adding test cases for >> these?

> > Yeah, this time I do :) > > (I added tests, but not for this specific case before) > > [snip]

>>> Clearly not optimal...

>> >> Yes. With my test code: >> >> #!/usr/local/bin/perl >> >> use strict; >> use Math::BigRat; >> >> my $rat = Math::BigRat->new( '67 / 100' ); >> my $exp = exp $rat; >> print $exp->as_float(15) , "\n"; >> >> # time perl exp.pl >> 1.95423732063594 >> 1.239u 0.029s 0:01.26 99.2% 10+1165k 0+0io 0pf+0w >> >> With 667/1000: >> >> # time perl exp.pl >> 1.94838339395450 >> 79.383u 0.037s 1:19.45 99.9% 10+1137k 0+0io 0pf+0w >> >> With 6667/10000, well, I got tired of waiting on it after 30 minutes. >> >> Is there anything that can be done to improve the performance?

> > Yeah I think so, but I need firt to think of an algorithm to calculate > the bpow() result without losing precision. > > Probably will release the next version just as it is, because I don't > know how long it will take to do this. > > Thanx for the report again, > > Tels > > -- > Signed on Sun Apr 6 21:27:57 2008 with key 0x93B84C15. > Get one of my photo posters: http://bloodgate.com/posters > PGP key on http://bloodgate.com/tels.asc or per email. > > "I have always wished that my computer would be as easy to use as my > telephone. My wish has come true. I no longer know how to use my > telephone." > > -- Bjarne Stroustrup, inventor of the C++ programming language > > <mime-attachment>

Mon Apr 07 15:15:38 2008 nospam-abuse [...] bloodgate.com - Correspondence added

Subject:	Re: [rt.cpan.org #34584] hang when calling exp
Date:	Mon, 7 Apr 2008 21:14:35 +0200
To:	bug-Math-BigRat [...] rt.cpan.org
From:	Tels <nospam-abuse [...] bloodgate.com>

Moin Steve, On Monday 07 April 2008 20:42:30 Steve Wills via RT wrote: Show quoted text

> Queue: Math-BigRat > Ticket <URL: http://rt.cpan.org/Ticket/Display.html?id=34584 > > > Hi there, > > Just wondering, when are you planning to release 0.22 to CPAN?

I just pushed Math::BigRat v0.22 to CPAN - it should propagate to the mirros in one hour or so. Since I can only work on these modules in my (little) spare time, there is not that much I can do in short time frames - work, real life and other stuff takes precedence. The current version contains the fixes I already mentioned, but not yet an algorithm to calculate broot() more efficiently. For this, I need a few days and some time to develop something and then implement it - this is not a 5 min affair. Sorry. All the best, Tels -- Signed on Mon Apr 7 21:12:16 2008 with key 0x93B84C15. Get one of my photo posters: http://bloodgate.com/posters PGP key on http://bloodgate.com/tels.asc or per email. "Don't worry about people stealing your ideas. If your ideas are any good, you'll have to ram them down people's throats." -- Howard Aiken

Download (untitled)
application/pgp-signature 481b

Message body not shown because it is not plain text.

Mon Apr 07 15:21:47 2008 steve [...] mouf.net - Correspondence added

Subject:	Re: [rt.cpan.org #34584] hang when calling exp
Date:	Mon, 7 Apr 2008 15:21:00 -0400
To:	bug-Math-BigRat [...] rt.cpan.org
From:	Steve Wills <steve [...] mouf.net>

On Apr 7, 2008, at 3:15 PM, nospam-abuse@bloodgate.com via RT wrote: Show quoted text

> > <URL: http://rt.cpan.org/Ticket/Display.html?id=34584 > > > Moin Steve, > > On Monday 07 April 2008 20:42:30 Steve Wills via RT wrote:

>> Queue: Math-BigRat >> Ticket <URL: http://rt.cpan.org/Ticket/Display.html?id=34584 > >> >> Hi there, >> >> Just wondering, when are you planning to release 0.22 to CPAN?

> > I just pushed Math::BigRat v0.22 to CPAN - it should propagate to the > mirros in one hour or so. >

Great, thanks! Show quoted text

> Since I can only work on these modules in my (little) spare time, > there > is not that much I can do in short time frames - work, real life and > other stuff takes precedence. > > The current version contains the fixes I already mentioned, but not > yet > an algorithm to calculate broot() more efficiently. For this, I need a > few days and some time to develop something and then implement it - > this is not a 5 min affair. Sorry.

Of course, I understand. Your efforts are greatly appreciated. I don't mean to rush you, I was merely asking, not demanding. If there's anything I can do to help, please let me know. I'm happy to help test, of course. Steve Steve

Sun Oct 14 17:33:40 2012 CHORNY [...] cpan.org - Fixed in 0.22 added