[Chicken-users] Bug in socket egg and patch
Hello, While writing a socket server in Chicken for kicks I think I run into a bug in the sockets egg where socket-accept would not handle the error generated by select() if the listener socket had been closed - the error is already handled in the egg when selecting for writing. I am not sure how the procedure for reporting bugs for eggs or submitting patches works and am still very new to Chicken, but the chicken-install utility has been very nice so far and has made it very easy to test modifications (with -R and just chicken-install to reinstall a local version). Thanks for that, for the useful implementation, and for the egg! I have attached a patch that I believe fixes the bug. Regards, -- Jonathan Chan j...@fastmail.fm --- socket.scm 2013-11-27 00:21:47.016722696 -0800 +++ socket.new.scm 2013-11-27 00:21:40.046604461 -0800 @@ -737,17 +737,21 @@ (let ((s (socket-fileno so)) (to (socket-accept-timeout))) (let restart () - (if (eq? 1 (select-for-read s)) - (let ((s (_accept s #f #f))) -(when (eq? -1 s) - (network-error/errno 'socket-accept could not accept from listener so)) -(let ((so (make-socket s (socket-family so) (socket-type so) (socket-protocol so - (unless (_make_socket_nonblocking s) -(network-error/errno 'socket-accept unable to set socket to non-blocking so)) - so)) - (begin + (let ((f (select-for-read s))) +(cond + ((eq? f -1) + (network-error/errno 'socket-connect select failed so)) + ((eq? f 1) + (let ((s (_accept s #f #f))) + (when (eq? -1 s) + (network-error/errno 'socket-accept could not accept from listener so)) + (let ((so (make-socket s (socket-family so) (socket-type so) (socket-protocol so + (unless (_make_socket_nonblocking s) + (network-error/errno 'socket-accept unable to set socket to non-blocking so)) + so))) + (else (block-for-timeout! 'socket-accept to s #:input) -(restart)) +(restart))) ;; Returns number of bytes received. If 0, and socket is sock/stream, peer has shut down his side. (define (socket-receive! so buf #!optional (start 0) (end #f) (flags 0)) ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
Re: [Chicken-users] Problem: Chicken-install OpenSSL / rand.h
Hm, guess I need to install openssl. Right now I thought that I only need to have the source files on the hard drive. source: https://github.com/simio/teve P: chicken-install fails to build the openssl egg. S: Make sure you have the OpenSSL libs and header files installed. If they have been stashed away into some obscure directory, you may need to pass their locations to the compiler and linker: # env CSC_OPTIONS='-I /foo/include/openssl -L /foo/lib' \ make install-eggs -mfv ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
Re: [Chicken-users] go routines for chicken
Hi Piotr, m...@freeshell.de writes: Is there anything comparable in Chicken Scheme? If not, how complicated would be to make such an implementation? I am trying to port Clojure's core.async module in my very limited spare hacking time to CHICKEN right now. This should be what you are looking for. I will announce it here once something usable comes from that. I understand that Chicken only offers very limited threading capability. Well, it's only limited in the amount of cores you can utilize (1) and what kinds of functions you can run in parallel (e.g. FFI calls usually block all threads). This is orthogonal to Go-style channels, though. Moritz ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
Re: [Chicken-users] Problem: Chicken-install OpenSSL / rand.h
Hi, On Wed, 27 Nov 2013 13:21:53 +0100 m...@freeshell.de wrote: guess I need to install openssl. Right now I thought that I only need to have the source files on the hard drive. Right, you need openssl installed. source: https://github.com/simio/teve P: chicken-install fails to build the openssl egg. S: Make sure you have the OpenSSL libs and header files installed. If they have been stashed away into some obscure directory, you may need to pass their locations to the compiler and linker: # env CSC_OPTIONS='-I /foo/include/openssl -L /foo/lib' \ make install-eggs Thanks for the link. I didn't know that project. Very interesting. Best wishes. Mario -- http://parenteses.org/mario ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
Re: [Chicken-users] go routines for chicken
John Cowan co...@mercury.ccil.org writes: The mailbox egg provides the equivalent of Go channels. That's not quite right, is it? E.g. it is not possible to multiplex across mailboxes (as with Go's select statement) or limit their size (at least not out of the box). Those are quite essential provisions for programming in a CSP style, at least to my (so far fairly incomplete) understanding. Moritz ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
Re: [Chicken-users] Any hope for new zmq bindings to version 3.2 or even 4.0?
Hi Matt and Kristian, Kristian Lein-Mathisen kristianl...@gmail.com writes: Moritz and I had some fun with zmq 3.2 in July. We didn't release our work, with the reason slipping my mind right now. I don't remember either :-) Matt, can you try whether it works for you or if you still run into trouble with that version? Moritz ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
Re: [Chicken-users] Bug in socket egg and patch
Thanks. I'll take a look this weekend. Jim On Nov 27, 2013, at 2:29, Jonathan Chan j...@fastmail.fm wrote: Hello, While writing a socket server in Chicken for kicks I think I run into a bug in the sockets egg where socket-accept would not handle the error generated by select() if the listener socket had been closed - the error is already handled in the egg when selecting for writing. I am not sure how the procedure for reporting bugs for eggs or submitting patches works and am still very new to Chicken, but the chicken-install utility has been very nice so far and has made it very easy to test modifications (with -R and just chicken-install to reinstall a local version). Thanks for that, for the useful implementation, and for the egg! I have attached a patch that I believe fixes the bug. Regards, -- Jonathan Chan j...@fastmail.fm handle-errors-in-socket-accept.patch ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
Re: [Chicken-users] Any hope for new zmq bindings to version 3.2 or even 4.0?
I'll test this out soon. Thanks! On Nov 27, 2013 5:50 AM, Moritz Heidkamp mor...@twoticketsplease.de wrote: Hi Matt and Kristian, Kristian Lein-Mathisen kristianl...@gmail.com writes: Moritz and I had some fun with zmq 3.2 in July. We didn't release our work, with the reason slipping my mind right now. I don't remember either :-) Matt, can you try whether it works for you or if you still run into trouble with that version? Moritz ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
Re: [Chicken-users] go routines for chicken
Hi, I am trying to port Clojure's core.async module in my very limited spare hacking time to CHICKEN right now. This should be what you are looking for. I will announce it here once something usable comes from that. That is great news. I will watch out for that egg, even if it might take some time hatch it. I understand that Chicken only offers very limited threading capability. Well, it's only limited in the amount of cores you can utilize (1) and what kinds of functions you can run in parallel (e.g. FFI calls usually block all threads). This is orthogonal to Go-style channels, though. I see. I am not knowledgable in the area compilers/threads/OS processes, let alone in the magic that makes fairly efficiant C code come out of chicken. However, I imagine the the limitation when it comes to core usage is linked to the transcription technique of scheme to C. Otherwise one could peek in the C source of the go routines and translate them to chicken, no? Cheers! Piotr ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
Re: [Chicken-users] go routines for chicken
On a sidenote: It seems that there is an analgon to go routines on Erlang, which can be accessed the LFE (Lisp Flavoured Erlang). However, that would require to learn the entire Erlang VM ecosystem. ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
Re: [Chicken-users] go routines for chicken
On 27/11/13, m...@freeshell.de wrote: On a sidenote: It seems that there is an analgon to go routines on Erlang, which can be accessed the LFE (Lisp Flavoured Erlang). However, that would require to learn the entire Erlang VM ecosystem. A Lisp Flavoured Erlang, what nice!. ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
Re: [Chicken-users] go routines for chicken
Am 27.11.2013 20:32, schrieb Hugo Arregui: On 27/11/13, m...@freeshell.de wrote: On a sidenote: It seems that there is an analgon to go routines on Erlang, which can be accessed the LFE (Lisp Flavoured Erlang). However, that would require to learn the entire Erlang VM ecosystem. A Lisp Flavoured Erlang, what nice!. Having implemented a language inspired by Erlang in LISP (Scheme that is) and in a byzantine fault tolerant way atop; I feel from skimming over the discussion that I more or less have seen those related problems in practice. However the code I wrote to cope with them is for sure not conforming to any pre-defined API. Without making any promises (((and having already droped the page titled walkthough from this go chan's docs, thus given that go is about as great a search term as scheme makes pretty sure that I will not really be able to find it back again))): could I you please send me pointers to what is considered the canonical documentation of features and requirements a go-channel has to solve? ((I have only so much screen-reading time left per day because of my health. I'd rather just use it to code ahead instead of digging through stuff from the net.)) However for the same health issues I have a little too much spare time :-/ And as indicated: chances are that all I have to do is adapt some already tested code to fit the API as per spec. let's try /Jörg ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
Re: [Chicken-users] go routines for chicken
Go Routines are much like Chicken's SRFI-18 threads, except that they can multiplex over multiple *real* threads if one should block. Go also enjoys a rather robust Channels system, which is sort of like Scheme's ports, only it's type-safe by design. Wiki actually has a nice and short break-down on this: https://en.wikipedia.org/wiki/Goroutine#Concurrency -Dan On Wed, Nov 27, 2013 at 1:22 PM, Jörg F. Wittenberger joerg.wittenber...@softeyes.net wrote: Am 27.11.2013 20:32, schrieb Hugo Arregui: On 27/11/13, m...@freeshell.de wrote: On a sidenote: It seems that there is an analgon to go routines on Erlang, which can be accessed the LFE (Lisp Flavoured Erlang). However, that would require to learn the entire Erlang VM ecosystem. A Lisp Flavoured Erlang, what nice!. Having implemented a language inspired by Erlang in LISP (Scheme that is) and in a byzantine fault tolerant way atop; I feel from skimming over the discussion that I more or less have seen those related problems in practice. However the code I wrote to cope with them is for sure not conforming to any pre-defined API. Without making any promises (((and having already droped the page titled walkthough from this go chan's docs, thus given that go is about as great a search term as scheme makes pretty sure that I will not really be able to find it back again))): could I you please send me pointers to what is considered the canonical documentation of features and requirements a go-channel has to solve? ((I have only so much screen-reading time left per day because of my health. I'd rather just use it to code ahead instead of digging through stuff from the net.)) However for the same health issues I have a little too much spare time :-/ And as indicated: chances are that all I have to do is adapt some already tested code to fit the API as per spec. let's try /Jörg ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
Re: [Chicken-users] go routines for chicken
Hugo Arregui scripsit: A Lisp Flavoured Erlang, what nice!. http://lfe.github.io/ There is also a Ruby Flavored Erlang called Elixir which has become quite popular: see http://elixir-lang.org. They are all compatible at the Erlang VM level. -- One Word to write them all, John Cowan co...@ccil.org One Access to find them, http://www.ccil.org/~cowan One Excel to count them all, And thus to Windows bind them.--Mike Champion ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
[Chicken-users] ann bitstring 1.33
Hello! Want to notify about new release 1.33 changes: fixed signed integers parsing optimized integers matching implemented endian attribute for floating point types ! Please note that now big-endian is defualt endianess for floating types, this can break or slowdonw your old code. ! Use host attribute to restore old behavior (VALUE float) - (VALUE float host) implemented boolean type (bitconstruct (#t boolean)) - #u8(1) (bitconstruct (#t 32 boolean big)) - #u8(0 0 0 1) (bitmatch \x00 [((B 8 boolean)) B]) - #f new procs available (bitstring-bit-set? bitstring bit-index) (bitstring-reverse bitstring chunk-size endian) (bitstring-not bitstring) ___ Chicken-users mailing list Chicken-users@nongnu.org https://lists.nongnu.org/mailman/listinfo/chicken-users
