Re: [pcre-dev] 7.4-RC1 test release

トップ ページ
このメッセージを削除
著者: Philip Hazel
日付:  
To: Sheri
CC: pcre-dev
題目: Re: [pcre-dev] 7.4-RC1 test release
On Thu, 13 Sep 2007, Sheri wrote:

> Chances are that if someone wants to alter the meaning of \R, they also
> want to alter the newline setting, but currently the internalized
> settings are mutually exclusive since each one is only available when
> first. Could you work out putting multiple keywords in the option
> setting (*parentheses), e.g., (*ANY BSR_UNICODE)? or how about:
> ((*ANY)(*BSR_UNICODE))


I did think of that! You can have (*ANY)(*BSR_ANYCRLF) at the start if
you want. Perhaps the wording of the doc isn't clear? You can have as
many of the "specials" at the start as you like.

> (btw what does "bsr" stand for?)


Backslash R.

> Because I used the new default, my pcretest results now vary for
> patterns that include \R
> I suggest incorporating explicit options on such patterns so user
> results will match. The tests with issues are test2, test5, test7 and test8.


Indeed, the tests won't all work if you fiddle with newlines too much.
The automatic testing function for Linux skips some of them because of
this. It is just too hard - that is, I don't have the time - to make it
all totally automatic just at the moment.

If you configure with default newlines, then test2 should run - and it
contains tests that internally ask for different newline options. It's
just if you build with non-defaults that the stuff in test2 (etc.) that
do not specify newline options goes wrong.

I am now going to look at the build issues.

Philip

--
Philip Hazel, University of Cambridge Computing Service.