<div dir="ltr"><div dir="ltr">On Fri, Jul 2, 2021 at 2:05 AM Patrick Sanan <<a href="mailto:patrick.sanan@gmail.com">patrick.sanan@gmail.com</a>> wrote:<br></div><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">As you mention in [4], the proximate cause of the configure failure is this link error [8]:<br></blockquote><div><br></div><div>That missing function was introduced in  GCC 7.0, and is there only for i686, not x86_64. This looks like a bad GCC install to me.</div><div><br></div><div>   Matt</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
Naively, that looks like a problem to be resolved at the level of the C++ compiler and MPI.<br>
<br>
Unless there are wrinkles of this build process that I don't understand (likely), this [6] looks non-standard to me:<br>
<br>
        includedir="${prefix}/include"<br>
        ...<br>
        ./configure --prefix=${prefix} \<br>
                ...<br>
                -with-mpi-include="${includedir}" \<br>
                ...<br>
<br>
<br>
Is it possible to configure using  --with-mpi-dir, instead of the separate --with-mpi-include and --with-mpi-lib commands? <br>
<br>
<br>
As an aside, maybe Satish can say more, but I'm not sure if it's advisable to override variables in the make command [7].<br>
<br>
[8]     <a href="https://gist.github.com/jkozdon/c161fb15f2df23c3fbc0a5a095887ef8#file-configure-log-L7795" rel="noreferrer" target="_blank">https://gist.github.com/jkozdon/c161fb15f2df23c3fbc0a5a095887ef8#file-configure-log-L7795</a><br>
[6]     <a href="https://gist.github.com/jkozdon/c161fb15f2df23c3fbc0a5a095887ef8#file-build_tarballs-jl-L45" rel="noreferrer" target="_blank">https://gist.github.com/jkozdon/c161fb15f2df23c3fbc0a5a095887ef8#file-build_tarballs-jl-L45</a><br>
[7]     <a href="https://gist.github.com/jkozdon/c161fb15f2df23c3fbc0a5a095887ef8#file-build_tarballs-jl-L55" rel="noreferrer" target="_blank">https://gist.github.com/jkozdon/c161fb15f2df23c3fbc0a5a095887ef8#file-build_tarballs-jl-L55</a><br>
<br>
<br>
> Am 02.07.2021 um 06:25 schrieb Kozdon, Jeremy (CIV) <<a href="mailto:jekozdon@nps.edu" target="_blank">jekozdon@nps.edu</a>>:<br>
> <br>
> I have been talking with Boris Kaus and Patrick Sanan about trying to revive the Julia PETSc interface wrappers. One of the first things to get going is to use Julia's binary builder [1] to wrap more scalar, real, and int type builds of the PETSc library; the current distribution is just Real, double, Int32. I've been working on a PR for this [2] but have been running into some build issues on some architectures [3].<br>
> <br>
> I doubt that anyone here is an expert with Julia's binary builder system, but I was wondering if anyone who is better with the PETSc build system can see anything obvious from the configure.log [4] that might help me sort out what's going on.<br>
> <br>
> This exact script worked on 2020-08-20 [5] to build the libraries, se something has obviously changed with either the Julia build system and/or one (or more!) of the dependency binaries.<br>
> <br>
> For those that don't know, Julia's binary builder system essentially allows users to download binaries directly from the web for any system that the Julia Programing language distributes binaries for. So a (desktop) user can get MPI, PETSc, etc. without the headache of having to build anything from scratch; obviously on clusters you would still want to use system MPIs and what not.<br>
> <br>
> ----<br>
> <br>
> [1] <a href="https://github.com/JuliaPackaging/BinaryBuilder.jl" rel="noreferrer" target="_blank">https://github.com/JuliaPackaging/BinaryBuilder.jl</a><br>
> [2] <a href="https://github.com/JuliaPackaging/Yggdrasil/pull/3249" rel="noreferrer" target="_blank">https://github.com/JuliaPackaging/Yggdrasil/pull/3249</a><br>
> [3] <a href="https://github.com/JuliaPackaging/Yggdrasil/pull/3249#issuecomment-872698681" rel="noreferrer" target="_blank">https://github.com/JuliaPackaging/Yggdrasil/pull/3249#issuecomment-872698681</a><br>
> [4] <a href="https://gist.github.com/jkozdon/c161fb15f2df23c3fbc0a5a095887ef8#file-configure-log" rel="noreferrer" target="_blank">https://gist.github.com/jkozdon/c161fb15f2df23c3fbc0a5a095887ef8#file-configure-log</a><br>
> [5] <a href="https://github.com/JuliaBinaryWrappers/PETSc_jll.jl/releases/tag/PETSc-v3.13.4%2B0" rel="noreferrer" target="_blank">https://github.com/JuliaBinaryWrappers/PETSc_jll.jl/releases/tag/PETSc-v3.13.4%2B0</a><br>
<br>
</blockquote></div><br clear="all"><div><br></div>-- <br><div dir="ltr" class="gmail_signature"><div dir="ltr"><div><div dir="ltr"><div><div dir="ltr"><div>What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.<br>-- Norbert Wiener</div><div><br></div><div><a href="http://www.cse.buffalo.edu/~knepley/" target="_blank">https://www.cse.buffalo.edu/~knepley/</a><br></div></div></div></div></div></div></div></div>