www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Missing things in Phobos

reply Steve Teale <steve.teale britseyeview.com> writes:
I'm sure that there's quite a long list, but two things I've bumped into 
with using the MySQL protocol are:

1) We don't have SHA1 - I know there are people working on this. I've 
done a version that is as close to the reference C implementation as I 
could make myself keep it. This may be useful as a check for others doing 
more efficient implementations. If you want it, please let me know.

Related issue is should we have std.digest instead of std.md5

2) We don't have an off-the-shelf Socket class for Unix Sockets

Steve
Oct 20 2011
next sibling parent Gor Gyolchanyan <gor.f.gyolchanyan gmail.com> writes:
I think the solution to the second problem (along with an enormous
range of other problems) would be having the =D8MQ's binding in Phobos.
http://www.zeromq.org/

On Thu, Oct 20, 2011 at 1:37 PM, Steve Teale
<steve.teale britseyeview.com> wrote:

 I'm sure that there's quite a long list, but two things I've bumped into
 with using the MySQL protocol are:

 1) We don't have SHA1 - I know there are people working on this. I've
 done a version that is as close to the reference C implementation as I
 could make myself keep it. This may be useful as a check for others doing
 more efficient implementations. If you want it, please let me know.

 Related issue is should we have std.digest instead of std.md5

 2) We don't have an off-the-shelf Socket class for Unix Sockets

 Steve
Oct 20 2011
prev sibling next sibling parent Piotr Szturmaj <bncrbme jadamspam.pl> writes:
Steve Teale wrote:

 I'm sure that there's quite a long list, but two things I've bumped into
 with using the MySQL protocol are:

 1) We don't have SHA1 - I know there are people working on this. I've
 done a version that is as close to the reference C implementation as I
 could make myself keep it. This may be useful as a check for others doing
 more efficient implementations. If you want it, please let me know.


https://github.com/pszturmaj/dmisc/blob/master/sha.d

It works for octet/byte hashing. It needs some polishing for handling 
per bit hashing.


 Related issue is should we have std.digest instead of std.md5


What about:

std.crypto.hash.md5
std.crypto.hash.sha
std.crypto.hash.whirlpool
std.crypto.hash.etc?

or std.crypt.* ?
Oct 20 2011
prev sibling next sibling parent Sean Kelly <sean invisibleduck.org> writes:
On Oct 20, 2011, at 2:37 AM, Steve Teale <steve.teale britseyeview.com> wrot=
e:

=20
 2) We don't have an off-the-shelf Socket class for Unix Sockets


The lack of networking support is my biggest problem with Phobos right now. I=
 wish I had the time to do something about it.=
Oct 20 2011
prev sibling next sibling parent reply Daniel Gibson <metalcaedes gmail.com> writes:
Am 20.10.2011 11:37, schrieb Steve Teale:

 2) We don't have an off-the-shelf Socket class for Unix Sockets


It shouldn't be too hard to add AF_UNIX support to std.socket.
(Derive something suitable from std.socket.Address and possibly 
std.socket.Socket)
In fact, I have some old untested code lying around that adds such an 
Address type: http://pastebin.com/gMxd4AqB

It's from D1 though and /may/ need minor changes to work with the 
current std.socket. (Also it may be buggy, as it's untested)

(IPv6 support is also missing from std.socket btw)


 Steve


Cheers,
- Daniel
Oct 20 2011
parent reply Daniel Gibson <metalcaedes gmail.com> writes:
Am 20.10.2011 23:44, schrieb Daniel Gibson:

 Am 20.10.2011 11:37, schrieb Steve Teale:

 2) We don't have an off-the-shelf Socket class for Unix Sockets


 It shouldn't be too hard to add AF_UNIX support to std.socket.
 (Derive something suitable from std.socket.Address and possibly
 std.socket.Socket)
 In fact, I have some old untested code lying around that adds such an
 Address type: http://pastebin.com/gMxd4AqB

 It's from D1 though and /may/ need minor changes to work with the
 current std.socket. (Also it may be buggy, as it's untested)


Ok, while having a quick look at this old code I spotted the following bugs:
1. Line 23: path[1] does not exist if path.length==1 => check length
2. After Line 29 should be a return (or else { ... } with ... being 
lines 31-36)

Seems like I was tired or drunk or something when I wrote that ;)

Fixed (still untested) version: http://pastebin.com/Vb5QqJwS


 (IPv6 support is also missing from std.socket btw)


Cheers,
- Daniel
Oct 20 2011
parent Daniel Gibson <metalcaedes gmail.com> writes:
Am 21.10.2011 00:12, schrieb Daniel Gibson:

 Am 20.10.2011 23:44, schrieb Daniel Gibson:

 Am 20.10.2011 11:37, schrieb Steve Teale:

 2) We don't have an off-the-shelf Socket class for Unix Sockets


 It shouldn't be too hard to add AF_UNIX support to std.socket.
 (Derive something suitable from std.socket.Address and possibly
 std.socket.Socket)
 In fact, I have some old untested code lying around that adds such an
 Address type: http://pastebin.com/gMxd4AqB

 It's from D1 though and /may/ need minor changes to work with the
 current std.socket. (Also it may be buggy, as it's untested)


 Ok, while having a quick look at this old code I spotted the following
 bugs:
 1. Line 23: path[1] does not exist if path.length==1 => check length


Got it wrong again, this should work:
   if(path[0] == '\0' && (path.length==1 || path[1] == '\0'))
   	return;

Need sleep.
Oct 20 2011
prev sibling parent "Regan Heath" <regan netmail.co.nz> writes:
On Thu, 20 Oct 2011 10:37:33 +0100, Steve Teale  
<steve.teale britseyeview.com> wrote:


 I'm sure that there's quite a long list, but two things I've bumped into
 with using the MySQL protocol are:

 1) We don't have SHA1 - I know there are people working on this. I've
 done a version that is as close to the reference C implementation as I
 could make myself keep it. This may be useful as a check for others doing
 more efficient implementations. If you want it, please let me know.


I have a complete set of hashing functions (the original code which became  
the modules now present in Tango) I just need to tidy them up, add  
documentation, etc..  I don't know when I will get around to it however.


 Related issue is should we have std.digest instead of std.md5


Agreed.  Or std.crypto.digest or similar if we want to expand into the  
other areas of cryptography.

-- 
Using Opera's revolutionary email client: http://www.opera.com/mail/
Oct 24 2011