World: r3wp

Why not put false at the bottom ...

or is it the readability thing?

undecided about that

yeah, that's it

if we can keep the awake handler at this size, i'm fine with false 
at the bottom

Mines a bit long ...

well, see you around

Thanks muchly.

ftp STOR now supported ..

Just wondering .. if we want a graphic file transfer progress meter, 
we need a way to tap into the event handler to catch the number of 
bytes being written/read ...

Also need a way to stop a file transfer that is in progress ...

Is there going to be an approved method of implementing these things?

Or do we hack our own and hope that someone comes along and unifies 
all the schemes

http://www.linkedin.com/ppl/webprofile?vmi=&id=1690833&pvs=pp&authToken=iSCS&authType=name&locale=en_US&trk=ppro_viewmore&lnk=vw_pprofile


Holger's linkedin details ... wonder if someone could persuade him 
to do SSL again but for R3 ?

Small optimization:


Graham, if you try to connect to a host with an ip as a string, then 
a DNS lookup is performed even if it's useless.

On the other side, if you provide an ip as a tuple, no lookup is 
performed and it's a straight connection.


So as an example, when you enter in PASV mode (ftp proto) try this 
instead.

data-address: to-tuple to-block form copy/part tmp 4

so I put the open port where?

you don't need to, the first attempt is the right one

oh .. yeah ... ok. Good point.

you will receive a connect event but no lookup

works

Uploaded latest version with this change, and the STOR command .. 
just in time for bed!

not much else to do .. just delete, create directory, rename .... 
and append to existing file

i just wake up :-)

oh .. and then create the higher abstraction so that it supports 
the R2 of of reading and writing files ...

I just sent you an awake handler!

Anyone else where interested in helping polishing up these schemes?

Graham - we have all world REBOL developers here ... 5-10 ... so 
basically yes, we are interested, just pressed for the time :-)

so the old read-io of R2 ... is it just now read in R3?  Or must 
I know more than that?  I am porting my product to R3, so I am trying 
to get that part figured.  It's almost running now, besides this 
(maybe...fingers crossed)

just read ...

prot-send.r is just a function and not a scheme.  But it relies on 
system objects such as email ...  I guess we can include these in 
the function until some decision is made about where they should 
go.

The lack of TLS and SSL is going to be a big problem since so many 
people now use secure protocols.

R2 2.7.7 has mention in notes about TLS SSL.  My question is what 
is the test case these were tried on?  I just want to verify they 
work in R2.

This is R3 group.

I know, but a good point for heading in R3...use a common test case 
between the 2.

It's unlikely that tests would be portable - the port model is completely 
different in R3.

Understand all that.  Just what was done in 2.7.7 to test it?  I 
have tried without success.

http://www.gnu.org/software/gnutls/
Lesser GPL ... can this work?

Are there any issues with the lesser GPL and the open host kit?

Or, would this be core RT stuff ...?

GNU TLS supports

Certificate types:	X.509, OPENPGP
Protocols:	TLS1.2, TLS1.1, TLS1.0, SSL3.0

Ciphers:	AES-256-CBC, AES-128-CBC, 3DES-CBC, CAMELLIA-128, CAMELLIA-256, 
ARCFOUR, ARCFOUR-40
MACs:	SHA512, SHA384, SHA256, SHA1, RMD160, MD5

Key exchange algorithms:	RSA, RSA-EXPORT, DHE-DSS, DHE-RSA, DHE-PSK, 
PSK, SRP, SRP-RSA, SRP-DSS, ANON-DH
Compression methods:	DEFLATE, LZO, NULL

Extensions:	Max record size, Cert Type (OpenPGP), Server Name, SRP, 
TLS/IA, Opaque PRF Input

SHA256 ...need this for Amazon signing ...

No GPL stuff can be included in the host kit.

Not even LGPL.

not even if you compile it in yourself??

Only if you write it yourself and relicense it. No point to including 
it otherwise. In theory you could dynamicly link LGPL or Classpath 
code to the host, but you can't staticly link or otherwise incorporate 
it and distribute it. If you can't distribute it, there's no point 
in using the code.

Stick with BSD or similar licenses, like OpenSSL or the Putty SSL 
stack.

putty is the best place to start work on adding SSL to R3.  its all 
there, in MIT it is 100% stand alone and the code is compact and 
clean.

The only tricks are to see if it has proper TLS support and whether 
it is suceptible to the recently discovered SSL/TLS protocol vulns.

Putty hasn't been updated in a while and never reached 1.0. Still, 
I agree that it is the best bet.

I wonder why Carl can't reuse the r2 code ...