Author: Sheri Date: To: pcre-dev Subject: Re: [pcre-dev] PCRE 8.20-RC1 is available - Please Test
On 9/14/2011 11:31 AM, Philip Hazel wrote: > On Wed, 14 Sep 2011, Herczeg Zoltán wrote:
>
>> You are right the RunTest.bat was not updated. The strange thing is
>> that its current state shouldn't call JIT at all since its support is
>> totally missing from RunTest.bat and the basic tests does not force
>> JIT compilation. This is true for forcing pcre_study. Could you send
>> me the differences as well?
> You should have received Sheri's message via me; I forwarded it. You're
> right about RunTest.bat; it shouldn't have triggered JIT at all. Only
> tests 14 and 15 have JIT options actually in the testinput files, and
> RunTest.bat doesn't at the moment call those tests. 14 should be run
> when JIT support is enabled, and 15 when JIT support is not enabled.
> This is a real mystery. As I have no access to any Windows machines, I
> can't really help any more with this.
>
> RunTest.bat just seems to run all the tests regardless, assuming that
> PCRE is configured with full support for everything (UTF-8, Unicode
> properties) and a link size of 2, whereas RunTest checks the
> configuration and skips non-relevant tests. Does this matter?
>
> Philip
>
I think what happens in RunTest script is multiple executions of each of
the 15 tests
In the interim it writes testtry for each test and compares it to the
relevant testoutput.
Its rather difficult to establish what went wrong if there is a bad compare.
I've made three versions of RunTest.bat (RunTestNormal.bat,
RunTestStudy.bat and RunTestJIT.bat). I also worked out testing for
utf-8, ucp and jit support to skip appropriate tests.
RunTestNormal and RunTestStudy report no compare differences. But when I
use my RunTestJIT.bat file, I get the same comparison differences in
tests 1, 2, 4 and 11 that I had seen using the sh script in Msys.
Possibly this is connected with default newline of ANYCRLF with JIT in
effect. It appears that \n is not being recognized as a newline.
From the Windows cmd console:
> pcretest PCRE version 8.20-RC1 2011-09-12
re> /^abc$/m data> abc\nzzz 0: abc
data>
re> /^abc$/mS+ data> abc\nzzz No match
data>
re> /^abc$/mS data> abc\nzzz No match
But if I restart pcretest: >
> pcretest PCRE version 8.20-RC1 2011-09-12
re> /^abc$/mS data> abc\nzzz 0: abc
It seems that the other valid ANYCRLF newlines are also failing to be
recognized with S+ (I tried \r and \r\n in the data portion, and got No
match.)