Hi,
We compiled the latest RegEx library on:
1. SunOS 5.8 Workshop 6: We used the sunpro.mak makefile.
Unfortunately, we received many error messages regarding conversions from
enum to int
and vice versa.
2. HP-UX 11.11 aCC 3.33: We used the generic.mak makeilfe. We got an
error regarding dynamic_cast from char*& to char*.
In both cases, we didn't run the configure script in libs/config,
because if we do, we get many compilation errors (probably because of the
#define-s
in the created user.hpp).
Are you aware of these problems? or are we doing something wrong?
Is it imperative that we run the configure script and not use the
default user.hpp?
On top of that, we managed to overcome the problems I mentioned by
simply specifing explicit conversion and casting, and we compiled RegEx on
Forte 6 with the 64-bit compilation flag. The compilation worked,
but unfortunately, the tests failed. (I mean, the compiled okay and run, but
they say
that tests.txt contains errors - something that didn't happen when
compiling the 32-bit version).
What can we do to make it work in 64-bits?
Thanks,
Yaron
"John Maddock"
does boost regex support 64-bit for SunOS 5.8, HP-UX 11i (native compilers), WIN (MSVC-6) ? we are using an old(?) version - regexpp version 3.12. Is it recommended to upgrade to the latest version ? Is it essential to upgrade to the latest version with relevance to 64-bit compilation ?
64-bit support should be fine in most recent versions (there was a 64-bit related bug in one release, but I forget which one now, it was some time ago now). However the compilers you mention are particularly broken when it come to compiling any kind of vaguely advanced C++ : The sun compiler should just about handle regex, as will MSVC 6, but HP aCC is just so broken we've never been able to port even the most basic Boost libraries to it, sorry... but see also http://lists.boost.org/MailArchives/boost/msg63570.php
John.
We compiled the latest RegEx library on:
1. SunOS 5.8 Workshop 6: We used the sunpro.mak makefile. Unfortunately, we received many error messages regarding conversions from enum to int and vice versa.
You shouldn't have done so, it's either a compiler bug or a library one, can you please let me have the error messages privately?
2. HP-UX 11.11 aCC 3.33: We used the generic.mak makeilfe. We got an error regarding dynamic_cast from char*& to char*.
There are no dynamic_cast's in the library - and a cast from char*& to char* is a legal (implicit) cast is it not? Looks like a compiler bug. Where did the error occur, and what was the exact message?
In both cases, we didn't run the configure script in libs/config, because if we do, we get many compilation errors (probably because of the #define-s in the created user.hpp).
Are you aware of these problems? or are we doing something wrong? Is it imperative that we run the configure script and not use the default user.hpp?
No, I wasn't aware of those issues. The configure script should work, and needs to be reported as a bug if it doesn't (what errors did it introduce?), but you should not need to run it.
On top of that, we managed to overcome the problems I mentioned by simply specifing explicit conversion and casting, and we compiled RegEx on Forte 6 with the 64-bit compilation flag. The compilation worked, but unfortunately, the tests failed. (I mean, the compiled okay and run, but they say that tests.txt contains errors - something that didn't happen when compiling the 32-bit version).
Can you let me have the errors please? John.
participants (2)
-
Itay Eliaz -
John Maddock