perl-5.26.0

[bdubbs@…]

New minor version.

[ken@…]

For some non-core modules, expect breakage (particularly in test suites).

See e.g. ​http://blogs.perl.org/users/todd_rinaldo/2017/05/reviewing-perl-5-in-inc-at-the-perl-toolchain-summit.html

[Armin]

There seems to be an xz compressed tarball available. Consider using it.

[Armin]

chapter 5 fix

[Armin]

Note that I could not build perl in chapter 5 without the patch above.

[ken@…]

Yeah, that matches what I used last night. I still don't understand why it is only needed for the chapter 5 build (I left it in for chapter 6, thinking from the perl bug report that it must be g++-7, but then I had an issue with the tests (no output to the log once the tests had started to run, until they finished) and tried building it on another gcc-7.1 system. Looks good, but an explanation of why we need it would be even more useful :)

[bdubbs@…]

For Chapter 5, I am planing on using:

sed -i '9751,$ d' regexec.c

That accomplishes the same thing as the patch. As far as I know right now, the change is not needed for Chapter 6, but that's not verified.

[Armin]

-Duseshrplib in chapter 6 is why it doesn't fail there. Thing is, when built without this, all loadable modules will be built statically and linked into perl executable. The issue arrises because both a module and core perl have a function with same name defined within their code.

[ken@…]

Replying to Krejzi:

-Duseshrplib in chapter 6 is why it doesn't fail there. Thing is, when built without this, all loadable modules will be built statically and linked into perl executable. The issue arrises because both a module and core perl have a function with same name defined within their code.

Thanks.

[bdubbs@…]

Fixed at revision 11255.