My searches for any existing publication of those code snippets didn't shake out, so I waited for the 5GB download of the docker .tar and pulled out the files that ended in .ipg. I'm very cognizant that's not the whole story, but that's what I had the energy to do for now. I really wanted to see the PDF one, since that's actually the heuristic I use for evaluating any such "I can describe binary files" framework because that file format is ... special<p><a href="https://gist.github.com/mdaniel/cdf52de6a8aa8982d591da82b160a229" rel="nofollow">https://gist.github.com/mdaniel/cdf52de6a8aa8982d591da82b160...</a><p><pre><code> tar -xOf ipg-pldi-ae.tar e4a01dbc5b413d9709f0cf716cdb848725893b5f97e0870e09fd83e16839dfad/layer.tar \
| tar -xvf - \
home/opam/pldi-ae/IPG/spec/dns/ipg/dns.ipg \
home/opam/pldi-ae/IPG/spec/elf/ipg/elf.ipg \
home/opam/pldi-ae/IPG/spec/elf/ipg/readelf.ipg \
home/opam/pldi-ae/IPG/spec/gif/ipg/gif.ipg \
home/opam/pldi-ae/IPG/spec/ipv4/ipg/ipv4.ipg \
home/opam/pldi-ae/IPG/spec/pe/ipg/pe.ipg \
home/opam/pldi-ae/IPG/spec/zip/blackbox/unzip.ipg \
home/opam/pldi-ae/IPG/spec/zip/ipg/zip.ipg
</code></pre>
because Gist wouldn't let me use "/" in the filenames, I just replaced them with _ after killing the home/opam part; I left the rest so hopefully they'll show up in search results since pldi-ae and IPG are pretty distinct