Re: [Monotone-devel] renaming monotone executable (again)

[Justin Patrin]

On 3/4/06, Richard Levitte - VMS Whacker <address@hidden> wrote:
> In message <address@hidden> on Sat, 4 Mar 2006 01:47:42 -0800, Nathaniel 
> Smith <address@hidden> said:
>
> njs> On Wed, Mar 01, 2006 at 12:17:15PM +0100, Richard Levitte - VMS Whacker 
> wrote:
> njs> > Uhmm, please no.  A dot file like that is a portability
> njs> > nightmare, trust me (I'm still planning on porting to VMS if I
> njs> > can, I just got to find the time to upgrade the C++ compiler).
> njs>
> njs> I trust you, but could you elaborate anyway? :-)
>
> Sure.  On VMS, at least with the file system called ODS-2, which is
> still very common, files have the following syntax, expressed as a
> regexp:
>
>     [A-Z0-9\$_---]{0,31}\.[A-Z0-9\$_---]{0,31}
>
> Directories *always* have the extension .DIR, without exception.
> Therefore, a directory name starting with a period is simply
> impossible.
>
> This means that in all places where we have to deal with a dotted file
> name for a directory, we will have to have a conditional to define the
> alternative for VMS.  It might seem simple enough today, since there
> are probably not many places where MT is currently defined, but
> experience has shown that as a piece of software matures, things like
> names tend to spread around, be used by scripts (it already happens,
> see monotone-import.pl), whatever, and for every such place, there
> will have to be a conditional that selects something like _MTN for VMS
> and .MTN for Unix.  It gets annoying after a while to say the least.
>
> njs> I had a vague impression that VMS had enough interesting
> njs> differences that one more thing like this would not make much of
> njs> a difference either way, but this is not really based on anything...
>
> There's a pretty good Unix compatibility library these days, so at a C
> level, it's not too hard these days.  I will have to figure out a
> different way to do configuration, but other than that...  However,
> file system differences are still something to be aware of, they don't
> go away so easily.
>
> Besides, there's different ways to see this.  Yeah, there may be
> enough interesting differences already, why add another one?
>

Am I just being dense or wouldn't a simple conditional #define fix this?

--
Justin Patrin