ASL fails to compile

[jalving]

I've been running into an issue compiling the latest ASL and I'm not sure where to report this. I attached the output I'm getting. This is with a previously working PIPS installation. It seems the wget is grabbing a new ASL that introduced a compile error.

Can anyone reproduce this error?

---

jordan@spartacus-III:/opt/PIPS2/ThirdPartyLibs/ASL$ sudo ./wgetASL.sh

Downloading the third party packages for PIPS-NLP:

Downloading Ampl Solver Library (ASL):

--2019-01-09 14:11:14-- http://www.ampl.com/netlib/ampl/solvers.tgz Resolving www.ampl.com (www.ampl.com)... 70.38.71.71 Connecting to www.ampl.com (www.ampl.com)|70.38.71.71|:80... connected. HTTP request sent, awaiting response... 301 Moved Permanently Location: https://ampl.com/netlib/ampl/solvers.tgz [following] --2019-01-09 14:11:15-- https://ampl.com/netlib/ampl/solvers.tgz Resolving ampl.com (ampl.com)... 70.38.71.71 Connecting to ampl.com (ampl.com)|70.38.71.71|:443... connected. HTTP request sent, awaiting response... 200 OK Length: 349689 (341K) [application/x-gzip] Saving to: ‘solvers.tar.gz’

solvers.tar.gz 100%[====================================================================================================================>] 341.49K 931KB/s in 0.4s

2019-01-09 14:11:15 (931 KB/s) - ‘solvers.tar.gz’ saved [349689/349689]

ASL: Download Successful.

ln: failed to create symbolic link './src/solvers': File exists Applying patch for #define filename in asl.h, which is incompatible with mpi.h. rm -f amplsolver.a *.o arith.h details.c float.h stdio1.h cc -c -O -DASL_NO_FPINITMT fpinit.c cat stdio1.h0 >stdio1.h cc -O arithchk.c fpinit.o -lm ||\ cc -DNO_LONG_LONG -O arithchk.c fpinit.o -lm ||\ cc -DNO_SSIZE_T -O arithchk.c fpinit.o -lm ||\ cc -DNO_LONG_LONG -DNO_SSIZE_T -O arithchk.c fpinit.o -lm ./a.out >arith.h rm -f a.out arithchk.o fpinit.o sed "s!System_details!uname -sm!" details.c0 >details.c cc -c -O funcadd0.c cc -c -O asldate.c atof.c auxinfo.c avltree.c b_search.c basename.c bscanf.c com2eval.c comeval.c con1ival.c con2ival.c con2val.c conadj.c conpval.c conscale.c conval.c degree.c derprop.c details.c dtoa1.c duthes.c dynlink.c f_read.c fg_read.c fg_write.c fgh_read.c fpecatch.c fpinit.c fullhes.c func_add.c funcadd1.c g_fmt.c genrowno.c getenv.c getstub.c htcl.c indic_cons.c jac0dim.c jac2dim.c jacdim.c jacinc.c jacinc1.c libnamsave.c mach.c mainexit.c mip_pri.c misc.c mpec_adj.c mqpcheckv.c mypow.c names.c nl_obj.c nqpcheck.c nqpcheckZ.c obj2val.c obj_adj.c objprec.c objconst.c objval.c objval.c op_type.c pfg_read.c pfghread.c printf.c pshvprod.c punknown.c qp_read.c qpcheck.c qpcheckZ.c qsortv.c readsol.c repwhere.c rops.c rops2.c sigcatch.c sos_add.c sphes.c sscanf.c stderr.c studchk0.c sufsos.c value.c writesol.c wrtsol.c ws_desc.c wsu_desc.c x2check.c xectim.c xp1known.c xp2known.c jac0dim.c: In function ‘jac0dim_ASL’: jac0dim.c:207:10: error: ‘Edaginfo {aka struct Edaginfo}’ has no member named ‘nlc0’ asl->i.nlc0 = nlc; ^ jac0dim.c:208:10: error: ‘Edaginfo {aka struct Edaginfo}’ has no member named ‘nlo0’ asl->i.nlo0 = nlo; ^ sphes.c: In function ‘saveog’: sphes.c:521:22: error: ‘Edaginfo {aka struct Edaginfo}’ has no member named ‘nlc0’ for(pe = p + asl->i.nlc0; p < pe; p++) ^ sphes.c:540:22: error: ‘Edaginfo {aka struct Edaginfo}’ has no member named ‘nlc0’ for(pe = p + asl->i.nlc0; p < pe; p++) ^ sphes.c: In function ‘restog’: sphes.c:567:22: error: ‘Edaginfo {aka struct Edaginfo}’ has no member named ‘nlc0’ for(pe = p + asl->i.nlc0; p < pe; p++) ^ sphes.c: In function ‘sphes_setup_ASL’: sphes.c:764:15: error: ‘Edaginfo {aka struct Edaginfo}’ has no member named ‘nlc0’ nlc0 = asl->i.nlc0; ^ sphes.c: In function ‘sphes_ASL’: sphes.c:1091:15: error: ‘Edaginfo {aka struct Edaginfo}’ has no member named ‘nlc0’ nlc0 = asl->i.nlc0; ^ sphes.c:1154:22: error: ‘Edaginfo {aka struct Edaginfo}’ has no member named ‘nlc0’ for(pe = p + asl->i.nlc0; p < pe; p++, y1++) ^ makefile:185: recipe for target 'amplsolver.a' failed make: *** [amplsolver.a] Error 1

---

[cnpetra]

that's because of some recent changes in the ASL. I decided to not support AMPL/ASL because I am not aware of anyone using it. I will be removing from the build...

[jalving]

It looks like the Ipopt build is pulling an old ASL v1.3.0. I could change that script if you want to keep supporting it. It's just 2 lines.

coinasl=1.3.0
if wget -O solvers.tar.gz https://github.com/ampl/mp/archive/$coinasl.tar.gz

[nychiang]

I think it is better to keep it if this just needs 2 lines modification. In addition, the Argonne folks would like to use it for the debug reason. I will come back to this in the beginning of Feb, and also add support to the users' dual initial values.

[cnpetra]

if that's so easy to fix, definitely go for it

[cnpetra]

@nychiang : someone else reported the bug (on a private channel). How's the fix going?

[jalving]

@cnpetra @nychiang Since it's an easy change, I can make the pull request. Should only take a few minutes.

[jalving]

See this pull request for the proposed fix: https://github.com/Argonne-National-Laboratory/PIPS/pull/48

Admittedly, I was having issues building Cbc when I tested the install. I had to remove the make flags to get it to work. Can anyone else do a clean build to test?

[nychiang]

Thanks Jordan. I will test it, on Thursday or Friday. Cheers

On Tue, Feb 5, 2019, 12:30 PM Jordan Jalving notifications@github.com wrote:

See this pull request for the proposed fix: #48 https://github.com/Argonne-National-Laboratory/PIPS/pull/48

Admittedly, I was having issues building Cbc when I tested the install. I had to remove the make flags to get it to work. Can anyone else do a clean build to test?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/Argonne-National-Laboratory/PIPS/issues/45#issuecomment-460728667, or mute the thread https://github.com/notifications/unsubscribe-auth/AJpqPJZVbjyZyGQ1hCZLhS9iNgg6F6x3ks5vKb-pgaJpZM4Z4PRS .

[cnpetra]

addressed by PR #48; it is merged in the master now, together with a fix for the mumps/parmetis build PR #46

[nychiang]

I did a clean build and it is totally fine. Cheers,

NY

On Tue, Feb 5, 2019, 1:33 PM Cosmin G Petra notifications@github.com wrote:

Closed #45 https://github.com/Argonne-National-Laboratory/PIPS/issues/45 .

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/Argonne-National-Laboratory/PIPS/issues/45#event-2119305691, or mute the thread https://github.com/notifications/unsubscribe-auth/AJpqPLXW7uJjP-L96aZqaSouTruxpI7qks5vKc6DgaJpZM4Z4PRS .