On 23 Oct 2000, Ben Collins-Sussman wrote:
> APR is our main portability layer; in theory, it should make
> Subversion just as portable as Apache. In theory. ;)
Lots of things are portable, "in theory". *grin*
> Can you give details? I'm not sure what's required to make
> automake/autoconf work under cygwin. (If I recall, cygwin is usually
> auto-detected as a valid `target' system, no?)
.. I'm away from my NT machine at the moment, but basically it
came down to the configure scripts failing to run properly because
the perl I was using was a non-cygwin perl.
After that was solved there were problems with the way that
the cygwin tools mapped drive names around, so the location
I was trying to build at "d:/binaries/subversion/" became
"/cydrive/d/binaries/subversion/" - this caused problems, but
I cannot remember the detail off hand.
(Basically I stopped here.)
> Assuming APR works on Win32, it *should* be trivial to port the
> Subversion client. (The filesystem, however, won't compile without
> Berkeley DB -- and I have no idea if DB has been ported to Win32.)
The berkeley DB will build and run happily on Windows, I'm
not honestly sure if the current/cutting edge stuff will work.
But I do remember building a version for use with Dsniff, (an
ethernet packet sniffer).
> The client uses the neon library (www.webdav.org/neon) to speak to
> Apache; so we can use any auth mechanisms that neon & apache are
> capable of. Greg Stein can speak more about this.
Okay, that sounds reasonable.
It just seems "obvious" to me that all modern network programs/
protocols should be moving away from plaintext authentification.
> (Btw, I use NT Emacs all the time. Great tool. :) )
I won't argue with you there ;)
Steve
---
http://www.steve.org.uk/
Received on Sat Oct 21 14:36:12 2006