World: r3wp

btw - Beer does not use http, no? It uses plain tcp IIRC ...

yes

Will this proxy support allow https?

imo not ... proxy is not about security imo. Https requires ssl inbuilt, 
and it is only in Command imo. But if you mean that security should 
not be luxury and it should be provided by default in all rebol versions, 
then I am with you :-)

Not sure about this.

If we have access to the win32 http api, we should get https as well...

and not require command

I agree, if you can't use plugin under HTTPS, then it's largely useless 
in secure environments.

win32 api https
 http://blogs.msdn.com/ie/archive/2005/10/31/487509.aspx

WinInet.dll offers a Win32 API for http, https, and ftp downloads 
combined with other API for caching and parsing.  It\u2019s a very 
popular binary, and in addition to being part of the IE platform, 
is widely used in Windows client applications for its Networking 
services.

this one?

Looks good.  Does cookie handling as well.

makes it easier for us to pretend to IE.

These proposed refinements would use WinInet, and consequently, support 
HTTPS.

But no promises that would be available in a free version of the 
plugin -- need to check with Carl on that.

Josh: I guess this could be plugin-only and either replace read-thru 
etc. with a native or provide new port schemes (probably the former 
is much easier). if for some reason it's not possible to make the 
native read-thru compatible with the old mezzanine one then there 
would need to be a new name I guess, but I'm not sure Carl would 
like that (no more cross-platform, although the plugin is already 
single platform rigth now so maybe this is not a big problem for 
r2 and can be solved in r3)

Gabriele, yes you're right, it's not a cross-platform solution. However, 
this is a Win32 problem by definition....to explain: We are talking 
about providing a solution for clients that use an auto-config proxy 
server with Internet Explorer. By definition, these are Win32 clients 
that only need a Win32 solution.

There seems to be something wrong with the Firefox version of plugin:
>> 128 * 1.7
== 128,.0
Can anyone confirm this problem?

(In other words the same IE and FF version of rebol dll doesn't behave 
the same here on my setup.)

Cyphre, I get the same answer on both:

>> 128 * 1.7
== 217.6


But that doesn't mean much, since I know I have the same version 
of viewdll.dll installed. You can verify the version # of the REBOL 
DLL in the plugin by going to the following page: http://www.rebol.net/plugin/tests/test.html.

The version # is displayed in that test script.

Stability: Has anyone found bugs with a clean install of the plugin, 
besides the Win2k bug?

not me

Josh: the point is not what this solves, the point is that the same 
script will not work across platforms if we end up using a different 
function for each platform to do the same thing. now, since read-thru 
already does something similar, it looks like a good target to me. 
i'm not sure about a refinement to read, since in that case i guess 
it would be a hack or you'd rather need a new port scheme.

file-requester works in a similar way, by having a native function 
and a cross-platform wrapper. and read-thru does caching, maybe read 
is a better place.

Gabriele, you're right. If we do this at all, whatever function we 
choose should have a fallback implementation on other platforms.

JoshM: This is really strange...I binary compared the"C:\WINDOWS\Downloaded 
Program Files\viewdll.dll and C:\Program Files\Mozilla Firefox\plugins\viewdll.dll 
and they are same although the result of the mutiplication with decimal 
number differs.

Can you please check this test page in both IE and FF? http://cyphre.mysteria.cz/plugin-test.html

(note: I'm using IE 6.0.2800.1106.xpsp1.020828-1920 and FF 1.5.0.4)

BTW both plugin versions were auto-installed from the page at rebol.com

Anyway, I'll try to delete both plugins and install again to see 
if this helps.

libc somehow diferent? rebol uses something system-dependent, which 
is why windows has this 1e-02 and llinux not. maybe firefox has an 
own dll in the path?

Ok, I removed both plugins (IE and FF), cleared both browser caches 
and did atumatic instalation from www.rebol.com. Both instalations 
went ok. Then trid to run the test link from above but no luck:
result in IE: http://cyphre.mysteria.cz/tests/plugin-test-ie.gif
result in FF: http://cyphre.mysteria.cz/tests/plugin-test-ff.gif

libc has also some locale-setting. maybe firefox sets that? oh, i 
see it has ",." inside, not only ","?

yes, the ",." is very strange.

I noticed it when colors of one application wen't wrong in FF but 
looked good in IE. It was because for example in IE:
>> maroon * 1.7
== 217.0.0
but in FF:
>> maroon * 1.7
== 128.0.0

tries to convert by system and then fix that e-02 based on strings? 
oops, i overlooked the result is completely messed up. you get

>> 128 * 1.7
== 128,.0

and my linux

>> 28 * 1.7
== 47.6

aargh, because i cant copypaste..

but still == 217.6

Do you have the latest plugin for Linux?

not plugin. AFAIK there is noen for linux.

mentioned linx because windows and linux format different anyway.

Ah I though you are testing it under linux plugin ;)

would like that a lot! but time-machine broken :(

It would be good if anyone under Windows can try the test the link 
I posted.

now windows. can confirm, ff shows 128 * 1.7 = 128,.0

could be local-dependend? Germans use "," by default, czech too?

I get 128.0 too

sorry, 128,.0 actually :-)

DOes not explain the 128 instead of 217. But had it once on linux, 
where an app outputted with printf and save was broken. turned out 
printf stored everything with "," but expected still "." on loading. 
so when i see now numbers, "," and "wrong", its my first thought. 
;)