On 11-Oct-16 15:21, Lincoln A Baxter wrote: > If you look at Sys::SigAction, you can see some samples of this kind of > testing in Makefile.PL. (it prints both warnings, but continues... with > some tests disabled, and it returns NO-Supported status that causes the > Smoke testers to mark it as N/A (Microsoft Windows). Thanks. It looks as though you die, so unless there's some handler, you don't continue:
if ( $^O =~ /MSWin32/ ) {
die qq{
OS unsupported
Sys::SigAction is not Supported on on $^O operating systems
...
I see other places where you just warn.
I read CPAN::Reporter to discover that 'OS unsupported' in the output is
what triggers NA.
Otherwise, not generating a Makefile or Build triggers 'discard'.
What are the conventions for deciding which to use?
I gather that NA will show up in the test matrix; discard won't. NA is
slightly misleading
as the OS is OK, but the OpenSSL package isn't. But it would give a
clue that some systems are
out of rev, which might be interesting.
As far as I can tell, discard doesn't get reported at all. Which is
less informative, but results in a clean test matrix.
> On Tue, 2016-10-11 at 14:27 -0400, yary wrote:
>> On Tue, Oct 11, 2016 at 12:36 PM, Joel Maslak <[email protected]>
>> wrote:
>>> They backport security patches from new versions to the old
>> version, but
>>> don't backport most features. Thus even though RHEL 5 machines
>> might be
>>> running ".9.8e" (hopefully -40), they will have the critical
>> security
>>> patches - even though OpenSSL officially doesn't have them in
>> .9.8e.
>>
>> Sounds like the answer is a Gnu Configure-like philosophy of testing
>> for the particular security issues with a small probe-script, to be
>> compiled a Makefile.PL time.
>>
>> -y
smime.p7s
Description: S/MIME Cryptographic Signature
