#3 initial aarch64 and e2k support patches

[Michael Shigorin]

Hello Allen,
I've got libEMF building on e2k architecture -- it took me to fake out a context structure alike to what's done in aarch64 patch though; attaching both done against 1.0.11 (with e2k patch depending on aarch64 one's context, ahem).
Looks like it works for me with chemtool's EMF export.
Thank you for the library!

[Allen Barnett]

Thanks! I will try to apply your patches in the not-too-distant future.

[Allen Barnett]

On Sat, Mar 28, 2020 at 2:35 PM Michael Shigorin mike@altlinux.org wrote:

On Sat, Mar 28, 2020 at 01:39:31PM -0400, Allen Barnett wrote:

I have committed your patches to the SVN repository. Please
check out the trunk and make sure it functions as you expect.

Well almost, autoreconf -fisv has masked the need to update
config/config.{guess,sub} from recent gnu-config so it has
these arches handy too; would be perfect if you updated them
from http://git.savannah.gnu.org/gitweb/?p=config.git

I've tested r90 tree on e2kv4 host (with autoreconf),
had to add explicit -std=c++11 as lcc 1.23 the host CXX
would stick to c++03 by default so it frowns upon nullptr
(they've bumped it in 1.24 but we haven't migrated yet,
and being explicit with standards can't hurt).

`make CXXFLAGS+=-std=c++11 all check' ran just fine.

It's funny but I don't have aarch64 shell handy at the
moment; the config.* issue should be common with e2k
but -std one should be masked by current gcc's default.

Thank you!

PS: looks like your reply didn't make it to the ticket itself:
https://sourceforge.net/p/libemf/patches/3/ -- just in case.

I'm planning on deprecating the autoconf build in favor of CMake. I find
the autoconf stuff just to hard to maintain.

Thanks,
Allen

[Allen Barnett]

On Sat, Mar 28, 2020 at 3:08 PM Michael Shigorin mike@altlinux.org wrote:

On Sat, Mar 28, 2020 at 06:54:27PM -0000, Allen Barnett wrote:

Well almost, autoreconf -fisv has masked the need to update
config/config.{guess,sub} from recent gnu-config so it has
these arches handy too; would be perfect if you updated them
from http://git.savannah.gnu.org/gitweb/?p=config.git
I'm planning on deprecating the autoconf build in favor of
CMake. I find the autoconf stuff just to hard to maintain.

Well if good ol' autocrap is still planned to exist in the next
release, please update those two files, it's just about copying
them from upstream tree.

I find cmake to be even worse than autoconf (its infamous
"No known features for CXX compiler" took enough hair
from my head already) but it's up to you indeed :-)

Truly a case of "pick your poison" :-)

[Allen Barnett]

On Sat, Mar 28, 2020 at 3:25 PM Allen Barnett allenbarnett5@gmail.com
wrote:

On Sat, Mar 28, 2020 at 3:08 PM Michael Shigorin mike@altlinux.org
wrote:

On Sat, Mar 28, 2020 at 06:54:27PM -0000, Allen Barnett wrote:

Well almost, autoreconf -fisv has masked the need to update
config/config.{guess,sub} from recent gnu-config so it has
these arches handy too; would be perfect if you updated them
from http://git.savannah.gnu.org/gitweb/?p=config.git
I'm planning on deprecating the autoconf build in favor of
CMake. I find the autoconf stuff just to hard to maintain.

Well if good ol' autocrap is still planned to exist in the next
release, please update those two files, it's just about copying
them from upstream tree.

I find cmake to be even worse than autoconf (its infamous
"No known features for CXX compiler" took enough hair
from my head already) but it's up to you indeed :-)

Truly a case of "pick your poison" :-)

I committed the updated config.sub and config.guess. If you're happy, I
will close ticket #3.

Thanks,
Allen

[Allen Barnett]

Hi Michael: I have committed your patches to the SVN repository. Please
check out the trunk and make sure it functions as you expect.
Thanks,
Allen

On Sun, Sep 22, 2019 at 10:43 AM Michael Shigorin gvy@users.sourceforge.net
wrote:

---

• [patches:#3] https://sourceforge.net/p/libemf/patches/3/ initial
  aarch64 and e2k support patches*

Status: open
Group: Unstable_(example)
Labels: arch
Created: Sun Sep 22, 2019 02:43 PM UTC by Michael Shigorin
Last Updated: Sun Sep 22, 2019 02:43 PM UTC
Owner: nobody
Attachments:

• libEMF-aarch64.patch
  https://sourceforge.net/p/libemf/patches/3/attachment/libEMF-aarch64.patch
  (2.4 kB; text/x-patch)
• libEMF-e2k.patch
  https://sourceforge.net/p/libemf/patches/3/attachment/libEMF-e2k.patch
  (2.4 kB; text/x-patch)

Hello Allen,
I've got libEMF building on e2k architecture -- it took me to fake out a
context structure alike to what's done in aarch64 patch though; attaching
both done against 1.0.11 (with e2k patch depending on aarch64 one's
context, ahem).
Looks like it works for me with chemtool's EMF export.
Thank you for the library!

---

Sent from sourceforge.net because you indicated interest in
https://sourceforge.net/p/libemf/patches/3/

To unsubscribe from further messages, please visit
https://sourceforge.net/auth/subscriptions/

[Allen Barnett]

Done. And thanks for your contribution!

On Sat, Mar 28, 2020 at 4:02 PM Michael Shigorin mike@altlinux.org wrote:

On Sat, Mar 28, 2020 at 07:56:33PM -0000, Allen Barnett wrote:

I committed the updated config.sub and config.guess.

svn up; ./configure; make CXXFLAGS+=-std=c++11 all check
went fine, thank you!

If you're happy, I will close ticket #3.

I was happy with your initial commit already :-)

--
---- WBR, Michael Shigorin / http://altlinux.org
------ http://opennet.ru / http://anna-news.info

[Michael Shigorin]

Thank you, libemf 1.0.12 builds fine on both aarch64 and e2k with no extra patches.
Just for the record: the pristine tarball builds on e2k with lcc 1.24.09 (while 1.23.21 would error out even with explicit -std=c++11 but that's another story apparently not worth solving).
PS: http://git.altlinux.org/tasks/250671