Quick answer is to test what type of OS you are compiling first thing in the
MAKE
and then move form there
We do this in dbd:Oracle MAKEFILE.PL
like this
my $os = $^O;
if ($os eq 'VMS') {
..bla bla
}
elsif (($os eq 'MSWin32') or ($os =~ /cygwin/i)) {
bla bla
} and so on
I am out of here for today but I will have some time on the morrow
Cheers
John Scoels
----- Original Message -----
From: "Martin J. Evans" <[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Tuesday, July 03, 2007 2:16 PM
Subject: Re: DBD test case conditional on compile option
John Scoles wrote:
I guess that counts as my first patch for DBD::ODBC lol
Want to go for a second one now then?
I don't need to do this but I'd like to. There are two C typedefs in two
separate header files and I'd like to protect against a compilation where
sizeof(type1) != sizeof(type2) - it's basically down to the fact that a
WCHAR in Windows being a 2 byte quantity and in UNIX the ODBC driver
managers can be built with WCHAR = 4 bytes or 2 bytes although by default
they are built with a WCHAR being 2 bytes. I'd like to make sure you
cannot build DBD::ODBC if sizeof(WCHAR) != 2 (because as it stands the
unicode I have will only work with 2 byte WCHAR.) You cannot use sizeof in
the C preprocessor as in:
#if (sizeof(WCHAR) != 2)
as far as I am aware. So the question becomes how can you test
sizeof(WCHAR) = 2 before building DBD::ODBC?
Martin
----- Original Message ----- From: "Martin Evans"
<[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Tuesday, July 03, 2007 12:56 PM
Subject: Re: DBD test case conditional on compile option
Thanks John,
Good idea, have implemented it as a dbh read-only attribute.
Martin
--
Martin J. Evans
Easysoft Limited
http://www.easysoft.com
John Scoles wrote:
You might want to add a property to the package DBD::ODBC::db called
can_unicode
set it during the Make process then read it in the test process??
We have a method like that in DBD::ORACLE
here is the stub (of course you know to change the ora_ )
sub ora_can_unicode {
my $dbh = shift;
my $refresh = shift;
# 0 = No Unicode support.
# 1 = National character set is Unicode-based.
# 2 = Database character set is Unicode-based.
# 3 = Both character sets are Unicode-based.
return $dbh->{ora_can_unicode}
if defined $dbh->{ora_can_unicode} && !$refresh;
my $nls = $dbh->ora_nls_parameters($refresh);
$dbh->{ora_can_unicode} = 0;
$dbh->{ora_can_unicode} += 1 if $nls->{NLS_NCHAR_CHARACTERSET} =~
/UTF/;
$dbh->{ora_can_unicode} += 2 if $nls->{NLS_CHARACTERSET} =~
/UTF/;
return $dbh->{ora_can_unicode};
}
----- Original Message ----- From: "Martin Evans"
<[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Tuesday, July 03, 2007 12:13 PM
Subject: DBD test case conditional on compile option
Hi,
I am modifying DBD::ODBC and a switch to perl Makefile.PL will cause
DBD::ODBC to be built with a different compile macro
(e.g. -DWITH_UNICODE). The problem is that some of the test cases are
only relevant if DBD::ODBC is compiled with this option.
Originally, I was going to modify Makefile.PL to create a file called
WITH_UNICODE and check for its existence in the test and skip the
tests if not found. However, I don't know how to remove the file in
the make clean rule of MakeMaker.
Any ideas how I can do this, or alternatives greatly appreciated.
Martin
--
Martin J. Evans
Easysoft Limited
http://www.easysoft.com
