Re: [racket-users] Bytecode problem with fxvector inside a macro

2019-05-08 Thread Matthew Flatt
At Thu, 9 May 2019 00:15:43 +0300, Dmitry Pavlov wrote: > > On 5/9/19 12:04 AM, Dmitry Pavlov wrote: > > Matthew, > > > > > >> The intended error here is "cannot marshal value that is embedded in > >> compiled code" at `raco make` time, because fxvectors are not supported > >> as literals. I'll

Re: [racket-users] Bytecode problem with fxvector inside a macro

2019-05-08 Thread Dmitry Pavlov
On 5/9/19 12:04 AM, Dmitry Pavlov wrote: Matthew, The intended error here is "cannot marshal value that is embedded in compiled code" at `raco make` time, because fxvectors are not supported as literals. I'll fix the bytecode writer to check for this case. OK, thank you. What would you

Re: [racket-users] Bytecode problem with fxvector inside a macro

2019-05-08 Thread Dmitry Pavlov
Matthew, The intended error here is "cannot marshal value that is embedded in compiled code" at `raco make` time, because fxvectors are not supported as literals. I'll fix the bytecode writer to check for this case. OK, thank you. What would you recommend, though, to users who want fxvectors

Re: [racket-users] Bytecode problem with fxvector inside a macro

2019-05-08 Thread Matthew Flatt
The intended error here is "cannot marshal value that is embedded in compiled code" at `raco make` time, because fxvectors are not supported as literals. I'll fix the bytecode writer to check for this case. Meanwhile, the fact that non-literal values can be coerced to syntax (as long as they

[racket-users] Bytecode problem with fxvector inside a macro

2019-05-08 Thread Dmitry Pavlov
Hello, I would like to report something that I see as inconsistent behavior of the bytecode compiler. The following short program (an artificial minimal reproducible example) works at first, but fails after raco make. My OS is Linux. $ cat one.rkt #lang racket (require (for-syntax