To get started with Gambit, I decided to try running the benchmark suite that was included in the tarball. I started with one of the examples:
./bench gambit "fib tak"
This ran fine and gave me some results. However, when I tried this:
./bench gambit-int "fib"
I got this:
*** ERROR IN "fib.scm"@60.1 -- Ill-placed 'define-macro' 0.01 real 0.01 user 0.00 sys
Also, when I tried running with a different benchmark setting, such as:
./bench -s fixflo gambit "fib"
I get another error, this time:
Running... *** WARNING -- Variable "fxring+" used in module "fib.o1" is undefined *** WARNING -- Variable "fxring-" used in module "fib.o1" is undefined *** ERROR IN fib -- Operator is not a PROCEDURE (#!unbound 35 1)
Can any of the experts provide some pointers for me here? I'd have more persistence in debugging my own code if I could get the benchmarks working. I'm running on a MacBook with OS X 10.4.8 if it matters.
thanks Shawn
Afficher les réponses par date
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 11-Jan-07, at 12:01 PM, Shawn Hyam wrote:
Can any of the experts provide some pointers for me here? I'd have more persistence in debugging my own code if I could get the benchmarks working. I'm running on a MacBook with OS X 10.4.8 if it matters.
Which version of Gambit are you using? I just tried beta 20 and at least one of the tests you claim does not work actually works with beta 20.
Marc
I'm also using beta 20, but I'm not surprised that the tests work for you. I'm just not sure where I've gone wrong... I was hoping one of the errors I got would help you point me at the problem. I'll re- install everything from scratch and see if I have better luck.
thanks Shawn
On 11-Jan-07, at 12:53 PM, Marc Feeley wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 11-Jan-07, at 12:01 PM, Shawn Hyam wrote:
Can any of the experts provide some pointers for me here? I'd have more persistence in debugging my own code if I could get the benchmarks working. I'm running on a MacBook with OS X 10.4.8 if it matters.
Which version of Gambit are you using? I just tried beta 20 and at least one of the tests you claim does not work actually works with beta 20.
Marc
-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (Darwin)
iD8DBQFFpnmb//V9Zc2T/v4RAhtTAKC/Pn07pKWaxzPJ9uzFQpbAQZ43PACeKxKy 1vJzXoOCmkzGSmzjuABo/5I= =x3xK -----END PGP SIGNATURE-----
Well, it seems that the problem is something to do with the "if- fixflo" macro, or the interaction between it and the macros in the prefix-gambit-int.scm file. If I comment out the use of this macro in the prefix file then things work fine -- but if it's there, I get this error:
*** ERROR IN "fib.scm"@60.1 -- Ill-placed 'define-macro'
I'm way out of my league with this stuff, so I'll submit it as a data point and continue getting familiar with Gambit. Despite my problems, it seems like a great system, btw!
Shawn
On 11-Jan-07, at 2:09 PM, Shawn Hyam wrote:
I'm also using beta 20, but I'm not surprised that the tests work for you. I'm just not sure where I've gone wrong... I was hoping one of the errors I got would help you point me at the problem. I'll re- install everything from scratch and see if I have better luck.
thanks Shawn
On 11-Jan-07, at 12:53 PM, Marc Feeley wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 11-Jan-07, at 12:01 PM, Shawn Hyam wrote:
Can any of the experts provide some pointers for me here? I'd have more persistence in debugging my own code if I could get the benchmarks working. I'm running on a MacBook with OS X 10.4.8 if it matters.
Which version of Gambit are you using? I just tried beta 20 and at least one of the tests you claim does not work actually works with beta 20.
Marc
-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (Darwin)
iD8DBQFFpnmb//V9Zc2T/v4RAhtTAKC/Pn07pKWaxzPJ9uzFQpbAQZ43PACeKxKy 1vJzXoOCmkzGSmzjuABo/5I= =x3xK -----END PGP SIGNATURE-----
Gambit-list mailing list Gambit-list@iro.umontreal.ca https://webmail.iro.umontreal.ca/mailman/listinfo/gambit-list
Another update: I recruited a friend to download Gambit and the benchmarks to his MacBook Pro, do a ./configure && make && make install and then try to run one of the benchmarks for gambit-int. Result: exactly the same error as I see.
Is there some configuration option missing, or maybe some code in ~/ gambcini or gambcext.scm that is required? Any other ideas? I am having a lot of trouble getting anything using macros to work properly.
Shawn
On 11-Jan-07, at 2:41 PM, Shawn Hyam wrote:
Well, it seems that the problem is something to do with the "if- fixflo" macro, or the interaction between it and the macros in the prefix-gambit-int.scm file. If I comment out the use of this macro in the prefix file then things work fine -- but if it's there, I get this error:
*** ERROR IN "fib.scm"@60.1 -- Ill-placed 'define-macro'
I'm way out of my league with this stuff, so I'll submit it as a data point and continue getting familiar with Gambit. Despite my problems, it seems like a great system, btw!
Shawn
On 11-Jan-07, at 2:09 PM, Shawn Hyam wrote:
I'm also using beta 20, but I'm not surprised that the tests work for you. I'm just not sure where I've gone wrong... I was hoping one of the errors I got would help you point me at the problem. I'll re- install everything from scratch and see if I have better luck.
thanks Shawn
On 11-Jan-07, at 12:53 PM, Marc Feeley wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 11-Jan-07, at 12:01 PM, Shawn Hyam wrote:
Can any of the experts provide some pointers for me here? I'd have more persistence in debugging my own code if I could get the benchmarks working. I'm running on a MacBook with OS X 10.4.8 if it matters.
Which version of Gambit are you using? I just tried beta 20 and at least one of the tests you claim does not work actually works with beta 20.
Marc
-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (Darwin)
iD8DBQFFpnmb//V9Zc2T/v4RAhtTAKC/Pn07pKWaxzPJ9uzFQpbAQZ43PACeKxKy 1vJzXoOCmkzGSmzjuABo/5I= =x3xK -----END PGP SIGNATURE-----
Gambit-list mailing list Gambit-list@iro.umontreal.ca https://webmail.iro.umontreal.ca/mailman/listinfo/gambit-list
Did you do a "make install" before running the benchmarks? I believe the benchmarks call the installed gsc/gsi, so if they're old then the benchmarks might not work.
Brad
-
Bradley Lucier -
Marc Feeley -
Shawn Hyam