At Fri, 27 Dec 2019 14:01:33 +0300, Dmitry Pavlov wrote:
> Because on my good ol' Windows XP machine, the Utah i386 nigtly build, while
> installing successfully, resulted in a Racket executable that the system
> refused to run, saying that it is not a correct executable. (The message was
>
Matthew,
Only today I managed to get back to that problem.
But your effort of making a special executable for me still turned out to be
relevant.
Because on my good ol' Windows XP machine, the Utah i386 nigtly build, while
installing successfully, resulted in a Racket executable that the system
At Sun, 22 Dec 2019 20:28:41 +0300, Dmitry Pavlov wrote:
>
> > Thanks! It really is a bug in `scheme_get_long_long_val`, where the
> > extraction can read past the end of a bignum on a 32-bit platform.
> >
> > Repair pushed.
> Great, thank you!
> Given that I do not normally build Racket, should
Thanks! It really is a bug in `scheme_get_long_long_val`, where the
extraction can read past the end of a bignum on a 32-bit platform.
Repair pushed.
Great, thank you!
Given that I do not normally build Racket, should I wait for the next snapshot
from UoU or Northwestern to check out the
At Sat, 21 Dec 2019 11:06:33 +0300, Dmitry Pavlov wrote:
> The error pops out not on the first or second positioning exceeding 1 GB. In
> fact, it happens on different moments in different runs.
Thanks! It really is a bug in `scheme_get_long_long_val`, where the
extraction can read past the end
At Sat, 21 Dec 2019 00:22:19 -0500, George Neuner wrote:
> Is Racket really writing billions of zeroes rather than creating a
> sparse file? It seems to me that this file should only create 2 actual
> data blocks, and (modulo JIT) the program should finish almost
> instantly. What am I
There could be something wrong with `scheme_get_long_long_val`, but
it's surprising that it could go wrong in a machine-specific way.
Does the error happen for you even in a very short program that tries
to set the file position to 1122398240, or does it only happen in your
full program?
I
On 12/20/2019 6:59 PM, Matthew Flatt wrote:
At Fri, 20 Dec 2019 23:39:30 +0300, Dmitry Pavlov wrote:
> > The Racket-imposed limit should be 64 bits (more than enough) on all
> > platforms. I can try to replicate the problem later today, but more
> > information on the error message would be
At Fri, 20 Dec 2019 23:39:30 +0300, Dmitry Pavlov wrote:
> > The Racket-imposed limit should be 64 bits (more than enough) on all
> > platforms. I can try to replicate the problem later today, but more
> > information on the error message would be helpful.
>
> I do not have access to that Windows
On 12/20/2019 2:49 PM, Dmitry Pavlov wrote:
I would have understood maximum the limit of maximum signed integer
2^31 = 2 GB (and my program would be fine with 2 GB limit).
meant to be "I would have understood the limit of maximum signed
integer ..."
The fixnum tag is just 1 bit, but you
The Racket-imposed limit should be 64 bits (more than enough) on all
platforms. I can try to replicate the problem later today, but more
information on the error message would be helpful.
I do not have access to that Windows 7 machine until Monday.
I managed to reproduce the problem, though,
On Fri, Dec 20, 2019 at 2:05 PM Shu-Hung You
wrote:
>
> Could it be that the Racket layer importing rktio C code directly uses
> get-ffi-obj and does not convert scheme_bignum_type to _int64?
>
Never mind this -- I got the FFI part wrong.
> On Fri, Dec 20, 2019 at 2:01 PM Matthew Flatt wrote:
Could it be that the Racket layer importing rktio C code directly uses
get-ffi-obj and does not convert scheme_bignum_type to _int64?
@Dmitry: I guess Racket uses 1 bit for type tagging, leaving only
30+sign (or 62+sign) bits for the actual value.
On Fri, Dec 20, 2019 at 2:01 PM Matthew Flatt
The Racket-imposed limit should be 64 bits (more than enough) on all
platforms. I can try to replicate the problem later today, but more
information on the error message would be helpful.
At Fri, 20 Dec 2019 17:39:37 +0300, Dmitry Pavlov wrote:
> Hello,
>
> On a fresh 32-bit Racket 7.5 install
I would have understood maximum the limit of maximum signed integer
2^31 = 2 GB (and my program would be fine with 2 GB limit).
meant to be "I would have understood the limit of maximum signed integer
..."
--
You received this message because you are subscribed to the Google Groups "Racket
On a fresh 32-bit Racket 7.5 install on 32-bit Windows 7,
(file-position port number) does not work when number
is more that 1 GB.
On 32-bit machines, the largest fixnum is 2^30 which is 1GB. I
suspect that /file-position/ really wants a fixnum rather than a
non-negative-integer as is
On 12/20/2019 9:39 AM, Dmitry Pavlov wrote:
On a fresh 32-bit Racket 7.5 install on 32-bit Windows 7,
(file-position port number) does not work when number
is more that 1 GB.
On 32-bit machines, the largest fixnum is 2^30 which is 1GB. I suspect
that /file-position/ really wants a
17 matches
Mail list logo
