Re: flex (was Re: [HACKERS] Installation procedure.) - Mailing list pgsql-hackers

From J. Michael Roberts
Subject Re: flex (was Re: [HACKERS] Installation procedure.)
Date
Msg-id Pine.GSO.4.10.9908021042230.1558-100000@school.cs.indiana.edu
Whole thread Raw
In response to flex (was Re: [HACKERS] Installation procedure.)  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: flex (was Re: [HACKERS] Installation procedure.)
List pgsql-hackers
> "J. Michael Roberts" <mirobert@cs.indiana.edu> writes:
> > - I needed to install flex (no surprise) -- the instructions are quite
> >   explicit, but, well, wrong: flex depends on bison.  So you have to get
> >   and compile bison first.
> 
> BTW, does anyone understand *why* our lexer files require flex and not
> just garden-variety lex?  Would it be worth trying to make them more
> portable?
> 
> Or perhaps we should ship pre-lexed derived files, as we do for the
> larger grammar files?
> 
> Having to install bison & flex is probably the most annoying Postgres
> prerequisite for people on non-Linux platforms, so I think it would
> be nice to clean this up.  I hadn't realized that you're essentially
> forced to install both...

You undoubtedly already had both installed...

For the record, if configure doesn't find flex, it assumes lex.  The
problem is if you don't even have lex.

However, the pre-lexed derived files are a good idea.  If somebody then
really wants to mess with those, they can go get flex.  That means that
the standard distribution *wouldn't* require flex.  For me, it was a good
excuse to get the lead out and finally install flex and bison (OK, it took
me only about fifteen minutes, but you know how those to-do lists get).
But if my only goal were just to get Postgres running, that would be a
rather unnecessary step.



pgsql-hackers by date:

Previous
From: The Hermit Hacker
Date:
Subject: ...
Next
From: Michael J Schout
Date:
Subject: bytea type and precision.