converting frobby into a new-style package

[dimpase]

currenly frobby is broken due to a C++ issue which Macaulay2 people know how to fix.

This ticket converts frobby into a new-style package and applies this fix, and other patches they came up with.

The upstream tarball is here: ​http://www.broune.com/frobby/frobby_v0.9.0.tar.gz

[dimpase]

New commits:

​1680e35

initial commit and a minimal patch, so that it builds

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​339252f

All M2 patches integrated, rpath to link correctly

[dimpase]

SPKG_CHECK=yes sage -i frobby returns OK, and all the tests in the frobby interface pass. There are some C++ issues in the debugging version of the program, but they can be ignored for our purposes.

[embray]

Does frobby have any dependencies that should be listed?

[dimpase]

Replying to embray:

Does frobby have any dependencies that should be listed?

it only depends on a standard Sage package (MPIR - which is impersonating GMP here).

[embray]

Probably still good to add a dependencies file for that. You can use $(MP_LIBRARY) for this.

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​f1f80e4

frobby needs mpir/gmp

[dimpase]

Replying to embray:

Probably still good to add a dependencies file for that. You can use $(MP_LIBRARY) for this.

done. Indeed, this seems to be the correct thing to do.

[mkoeppe]

Upstream tarball = ?

[mkoeppe]

sage -f -c frobby gives:

[frobby-0.9.0.p2] Running the test suite for frobby-0.9.0.p2...
[frobby-0.9.0.p2] g++   -Wall -I /Users/mkoeppe/cvs/sage/local/include -Wno-uninitialized -Wno-unused-parameter -std=c++11 -g -D DEBUG -fno-inline -Wextra -Wno-uninitialized -Wno-unused-parameter -c src/main.cpp -o bin/debug/main.o
[frobby-0.9.0.p2] In file included from /Users/mkoeppe/cvs/sage/local/include/c++/4.9.2/map:60:0,
[frobby-0.9.0.p2]                  from src/CliParams.h:24,
[frobby-0.9.0.p2]                  from src/Action.h:21,
[frobby-0.9.0.p2]                  from src/main.cpp:20:
[frobby-0.9.0.p2] /Users/mkoeppe/cvs/sage/local/include/c++/4.9.2/bits/stl_tree.h: In member function 'std::_Rb_tree_node<_Val>* std::_Rb_tree<_Key, _Val, _KeyOfValue, _Compare, _Alloc>::_M_create_node(_Args&& ...)':
[frobby-0.9.0.p2] /Users/mkoeppe/cvs/sage/local/include/c++/4.9.2/bits/stl_tree.h:420:14: error: '__tmp' does not name a type
[frobby-0.9.0.p2]         ::new(__tmp) _Rb_tree_node<_Val>;
[frobby-0.9.0.p2]               ^
[frobby-0.9.0.p2] make[2]: *** [bin/debug/main.o] Error 1

but then it happily continues ("All normal tests passed.")

[mkoeppe]

Also, there is the optional frobgrob script, which depends on 4ti2. See the manual

[dimpase]

Replying to mkoeppe:

sage -f -c frobby gives:

[frobby-0.9.0.p2] Running the test suite for frobby-0.9.0.p2...
[frobby-0.9.0.p2] g++   -Wall -I /Users/mkoeppe/cvs/sage/local/include -Wno-uninitialized -Wno-unused-parameter -std=c++11 -g -D DEBUG -fno-inline -Wextra -Wno-uninitialized -Wno-unused-parameter -c src/main.cpp -o bin/debug/main.o
[frobby-0.9.0.p2] In file included from /Users/mkoeppe/cvs/sage/local/include/c++/4.9.2/map:60:0,
[frobby-0.9.0.p2]                  from src/CliParams.h:24,
[frobby-0.9.0.p2]                  from src/Action.h:21,
[frobby-0.9.0.p2]                  from src/main.cpp:20:
[frobby-0.9.0.p2] /Users/mkoeppe/cvs/sage/local/include/c++/4.9.2/bits/stl_tree.h: In member function 'std::_Rb_tree_node<_Val>* std::_Rb_tree<_Key, _Val, _KeyOfValue, _Compare, _Alloc>::_M_create_node(_Args&& ...)':
[frobby-0.9.0.p2] /Users/mkoeppe/cvs/sage/local/include/c++/4.9.2/bits/stl_tree.h:420:14: error: '__tmp' does not name a type
[frobby-0.9.0.p2]         ::new(__tmp) _Rb_tree_node<_Val>;
[frobby-0.9.0.p2]               ^
[frobby-0.9.0.p2] make[2]: *** [bin/debug/main.o] Error 1

but then it happily continues ("All normal tests passed.")

yes, I have seen this. This concerns debugging version of the program only, IMHO. My C++ is too rusty to get any idea what goes wrong there, besides vague recollections of "iterator traits" etc I got nightmares about... Hopefully it's OK as it is.

[dimpase]

Replying to mkoeppe:

Also, there is the optional frobgrob script, which depends on 4ti2. See the manual

should 4ti2 be in frobby dependencies?

[mkoeppe]

Replying to dimpase:

Replying to mkoeppe:

Also, there is the optional frobgrob script, which depends on 4ti2. See the manual

should 4ti2 be in frobby dependencies?

If I understand the manual correctly, 4ti2 is only used by the frobgrob script, which is currently not installed. I think it should be installed; and 4ti2 should be a "recommended package" rather than a real dependency (like those many packages that one "should" install before polymake).

[mkoeppe]

(However, Debian for example does not install "frobgrob".)

[tscrim]

Judging only from the warning, I think the warning is coming from the fact that the forward declaration of new in the header has an input parameter __tmp that not been given an explicit type. Just add the type given in the implementation to the header and it should go away.

[mkoeppe]

Follow-up (frobgrob): #21004.

[vbraun]

Reviewer name