Autor: Philip Hazel Data: A: Bob Rossi CC: pcre-dev Assumpte: Re: [pcre-dev] Reply to much discussion
On Thu, 22 Feb 2007, Bob Rossi wrote:
> Perfect! I'll test patch 7 as well.
I've done a number of tests of both patch 6 and patch 7:
Patch 6:
Couldn't find anything to comment on. The failures of pcregrep with a
non-NL newline are OK in the sense that they are not a problem of the
build process. They are a failure of the testing mechanism. Moving
forward to using ./pcretest -C to control these tests should help. Don't
worry about these. I will fix them up when I work on this stuff.
Patch 7:
The only problems I had I've already noted in a previous message:
* --enable-stack-for-recursion MUST be "on" by default. Out of interest,
I ran "time make check" for a default configuration with this option
both on and off. The real/user/sys times for "off" were 8.01/6.36/1.73
whereas for "on" they were 5.62/3.97/1.70. The "off" setting is for
environments where there is very limited stack - not typically a
problem for Unix.
* I want to keep the old --enable-newline-is-xxx settings for backwards
compatibility.
Daniel, do you want to fix those issues and make another patch, or shall
I just take it from here and learn how to change them myself?
To avoid more confusion, I'm not going to proceed until I hear "over to
you" from both Bob and Daniel.
When I do, as well as any minor code changes, I will also make a sweep
through the administrative documentation.
--
Philip Hazel, University of Cambridge Computing Service.