Bug with makesetup on FreeBSD

[Andrew Csillag]

Randall Hopper wrote:
> 
> Andrew Csillag:
> |Randall Hopper wrote:
> |> Andrew Csillag:
> |>  |makesetup in Python 1.5.1 and 1.5.2 bombs on lines in the Setup file
> |>  |that use backslash continuation to break a module spec across lines on
> |>  |FreeBSD.
> |>
> |> BTW FWIW, I just built 1.5.2 last night on 3.0-RELEASE using the 1.5.2c1
> |> port.  Worked fine.  But it may not invoke makesetup under the hood.
> |
> |It does invoke makesetup (that's how the Makefile in Modules gets
> |written).  I'm also running FreeBSD 2.2.8, so it may be a bug in /bin/sh
> |that has been subsequently fixed...  The quick test is to try this on
> |your 3.0 machine
> |
> |$ read line
> |some text here\
> |
> |On my 2.2.8 machine after I hit return after the \, I get a command line
> |prompt, not a "blank prompt" that would mean that the read wasn't done.
> 
> It must be something else then, because here with stock Bourne shell:
> 
>  |$ read line
>  |some text here\
>  |$ echo $line
>  |some text here\
> 
> I get the same behavior you describe, but no build breakage.
> 
> Randall

I figured it out!  If you build out of the ports tree, it's Tkinter
configuration is all on one line (see
/usr/ports/lang/python/files/Setup.tk), not broken by using backslash
continuation as is in the distributed Setup file.  I was building from
the source release from www.python.org, not from the ports tree as you
did, hence why you didn't run into it and I did.
-- 
"There are two major products that come out of Berkeley: 
LSD and UNIX. We don't believe this to be a coincidence." 
- Jeremy S. Anderson