spkg-configure.m4 for eclib

[dimpase]

the main issue here is how to check that the version installed on the system is good enough.

We check for a header that only appeared in 2018, that's all we could find atm.

[dimpase]

while at it, we should fix eclib's spkg-install, which should always have --with-flint, otherwise flint is not linked it (causing worse performance I guess).

[dimpase]

see ​https://github.com/JohnCremona/eclib/issues/56 for the version issue

[dimpase]

New commits:

​2f45c9e	spkg-configure for pari
​b622942	spkg-configure for pari_elldata
​2e11aed	spkg-configure's for PARI packages
​183f47c	check for PARI packages in PARI's spkg-configure
​72f3c20	spkg-configure for gp2c
​6eac550	as pari might become dummy, test with GAP
​ac207d5	don't export GP_DATA_DIR, remove other, unneeded GP* vars
​317f090	proper dependency resolution (all tests in PARI spkg-config)
​2c26115	docbuild workaround for external PARI
​0c8bf98	spkg-configure for eclib

[git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

​e53d6a7	spkg-configure for eclib
​6745e5d	use internal or external flint, always

[dimpase]

The current Debian eclib is too old, one needs to rule it out using something like ​https://github.com/JohnCremona/eclib/issues/56#issuecomment-519813168

---

New commits:

​e53d6a7	spkg-configure for eclib
​6745e5d	use internal or external flint, always

[git]

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

​63ead07

added test for versions >= v20190226

[dimpase]

To test this, I checked that eclib from Debian 10 does not get picked up (as it should be). Then I uninstalled the system's eclib and installed the current eclib into /usr/local (with the configure parameters as in spkg-install), and it got picked up just fine.

[isuruf]

configure:12023: /projects/66d93023-00f0-4c12-8a25-5d6d4e486740/sage-build/bin/x86_64-conda_cos6-linux-gnu-c++ -std=gnu++11 -o conftest -fvisibility-inlines-hidden -std=c++17 -fmessage-length=0 -march=nocona -mtune=haswell -ftree-vectorize -fPIC -fstack-protector-strong -fno-plt -O2 -ffunction-sections -pipe -I/projects/66d93023-00f0-4c12-8a25-5d6d4e486740/sage-build/include -DNDEBUG -D_FORTIFY_SOURCE=2 -O2 -I/projects/66d93023-00f0-4c12-8a25-5d6d4e486740/sage-build/include -Wl,-O2 -Wl,--sort-common -Wl,--as-needed -Wl,-z,relro -Wl,-z,now -Wl,--disable-new-dtags -Wl,--gc-sections -L/projects/66d93023-00f0-4c12-8a25-5d6d4e486740/sage-build/lib -Wl,-rpath,/projects/66d93023-00f0-4c12-8a25-5d6d4e486740/sage-build/lib -lec conftest.cpp -lmpfr -lgmp -lm  -lntl >&5
/projects/66d93023-00f0-4c12-8a25-5d6d4e486740/sage-build/bin/../lib/gcc/x86_64-conda_cos6-linux-gnu/7.3.0/../../../../x86_64-conda_cos6-linux-gnu/bin/ld: /tmp/ccEA25VC.o: in function `main':
conftest.cpp:(.text.startup.main+0x11): undefined reference to `show_version()'

Note that -lec comes before conftest.cpp and not after like the other libs.

[git]

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

​c4665b3

correct LIBS and LDFLAGS to link correctly

[dimpase]

OK, I've fixed this, it was a really bad bug there, not the least wrong number of parameters used for AC_RUN_IFELSE.

[git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

​dd2b3fa	redirect gp errors to config.log
​23f8448	spkg-configure for eclib
​af16082	use internal or external flint, always
​7a8d11e	added test for versions >= v20190226
​699f220	correct LIBS and LDFLAGS to link correctly

[dimpase]

rebased over updated #28242

[isuruf]

Works for me in conda