

        <cite>Ben Deane</cite>
      <time>0:00</time>
      <p>
          Point   free.
      </p>

        <cite>Conor</cite>
      <time>0:01</time>
      <p>
          Should   we?   I   mean   I   didn&#39;t   delete   it.   We&#39;ll   actually   take   before   we   get   to   the   profiling,   let&#39;s   take   a   we&#39;re   gonna   take   a   small   interlude   here.   Welcome   to   ADSP   the   podcast,   episode   76,   recorded   on   April   19th,   2022.   My   name   is   Connor,   and   today   with   my   co-host   Bryce,   we   continue   part   two   of   our   four-part   interview   with   Ben   Dean   on   C   algorithm.   One   of   the   things   too   that   um   in   in   going   over   uh   both   the   blog   that   I   still   have   not   gone   back   to   find   that   individual&#39;s   name,   but   if   you&#39;re   listening,   uh   your   name   will   come   up   in   either   this   episode   or   part   two   of   this   episode   once   I   get   to.   Uh   actually   we   can   just   go   find   it   right   now.
      </p>

        <cite>Bryce</cite>
      <time>0:53</time>
      <p>
          Because   that&#39;s   great   podcasting   serving   the   internet   live.
      </p>

        <cite>Conor</cite>
      <time>0:58</time>
      <p>
          Here   we   go.   Felipe,   I   wrote   the   solutions   you   guys   discussed   in   this   episode.   I   I&#39;d   appreciate   if   you   could   have   a   look   at   the   reduce   slash   transform   reduce   solutions.   And   let   me   know   if   this   is   what   you   had   in   mind.   Love   the   episode.   So   thank   you,   Felipe.   I   responded   and   yeah,   it   looked   good.   I   only   perused   it   quickly.   He   was   missing   one   of   the   function   objects   or   lambdas   called   max2   underscore   accumulating,   but   other   than   that,   it   looked   good.
      </p>

        <cite>Bryce</cite>
      <time>1:19</time>
      <p>
          This   podcast   is   now   so   big   that   people   write   commentary   on   the   podcast.
      </p>

        <cite>Conor</cite>
      <time>1:25</time>
      <p>
          It&#39;s   not   that   big.   Uh   I   was   talking   to   Jason   the   other   day   at   the   Denver   C   meetup,   and   ours   is   uh   not   an   order   of   magnitude   smaller,   but   we   got   a   long   way   to   go   uh   to   call   it   big   as   being   generous.
      </p>

        <cite>Bryce</cite>
      <time>1:36</time>
      <p>
          Um   I   need   swag   for   C   now.   Do   we   have   stickers?   Do   we   have   a   logo?
      </p>

        <cite>Conor</cite>
      <time>1:42</time>
      <p>
          No,   we   put   and   here&#39;s   the   thing.   Uh   we   one   time   got   a   request   for   uh   transcripts.   Um,   but   we   have   to   understand   is   we   will   get   transcripts   as   soon   as   some   company   offers   uh   to   pay,   but   it&#39;s   not   cheap   to   because   like   I   do   not   have   the   bandwidth   to   uh   go   and   you   know   have   them   automatically   generated   and   then   clean   it   up.   Um   and   so   the   alternative   is   and   I   I   know   Ben   uh   Babryce   doesn&#39;t   have   the   bandwidth   either,   and   neither   does   Ben,   and   neither   would   he   be   interested   in   doing   that.
      </p>

        <cite>Ben Deane</cite>
      <time>2:10</time>
      <p>
          Um   but   so   you   have   to   get-   I   did   that   once.   I   think   it   was   the   Constexpra   All   the   Things   talk   that   Jason   and   I   did,   and   I   think   that   year   um   there   was   a   push   to   get   transcripts.   Um   and   you   know,   Google,   YouTube   does   automatic   kind   of   transcription   and   just   gets   it   horrendously   wrong   when   it   comes   to   C   keywords.   Constexpra,   you   know,   that&#39;s   not   a   word   that   Google   transcription   software   has   heard   of,   at   least   not   in   2017.   So   that   was   an   interesting   fix-up.
      </p>

        <cite>Conor</cite>
      <time>2:41</time>
      <p>
          And   so   that&#39;s   that&#39;s   the   thing.   Bryce   and   I   uh   we   do   really   care   a   ton   about   accessibility.   It&#39;s   just   that   it   is   not   it&#39;s   like   they   charge   by   the   minute   for   transcribing   that   stuff.   And   uh   so   if   there&#39;s   some   company   out   here   that&#39;s   listening   that&#39;s   like,   hey,   I   I   want   to   support   that   and   they   want   to   pay   for   that   to   get   done,   boom,   we&#39;re   gonna   have   transcripts.   But   I&#39;ve   already   got   like   16   things   on   the   go,   so   it&#39;s   it&#39;s   not   that   we   don&#39;t   care,   it&#39;s   that   it&#39;s   that   we&#39;re   cheap.   It&#39;s   prohibitive   to   do.   Don&#39;t   have   the   resources.   Yeah.   Um,   and   also   too,   like   if   I   had   the   bandwidth,   I   definitely   would.   I   just   I   currently   don&#39;t.   But   back   to   my   question.   At   this   point,   we   might   be   on   part   two   of   this   podcast.   Um,   is   that   one   of   the   things   that   really   led   me   to   prefer   uh   the   reduced   version   or   the   accumulate   version,   um,   and   something   we   didn&#39;t   really   touch   on   in   the   previous   uh   version   is   that   uh   or   maybe   I&#39;ll   ask   it.   I&#39;ll   ask   it   in   the   form   of   a   question.   An   interview   question.   I   was   trying   to   think   of   what   game   show   I   could   do,   but   I   don&#39;t   have   the   energy   to   think   of   a   game   show   and   put   it   in   the   right   format.   Um   sort,   partial   sort,   nth   element   all   have   something   in   common   that   transform   reduce   and   reduce   uh   and   accumulate   do   not,   which   makes   me   prefer   those.
      </p>

        <cite>Bryce</cite>
      <time>3:47</time>
      <p>
          Well,   sort   at   least   requires   random   access   iterators.   Is   that   true   of   the   others?
      </p>

        <cite>Conor</cite>
      <time>3:51</time>
      <p>
          Uh   not   what   I&#39;m   thinking   of,   even   if   it   is   true.
      </p>

        <cite>Bryce</cite>
      <time>3:53</time>
      <p>
          It   is   I   believe   it   is   true   of   all   of   them.   That&#39;s   a   bold   requirement   to   place   upon.
      </p>

        <cite>Conor</cite>
      <time>3:58</time>
      <p>
          So   to   repeat   the   repeat   the   question   to   the   listener   partial   sort,   sort,   and   nth   element   all   are   in   one   category,   reduce,   accumulate,   and   transform   reduce   are   all   in   another   category.   What&#39;s   the   difference?   That   I&#39;m   thinking   of.   There   might   be   multiple.
      </p>

        <cite>Ben Deane</cite>
      <time>4:11</time>
      <p>
          Yeah,   I   am   correct.   Is   it   the   header   therein?   Algorithm   versus   numeric.
      </p>

        <cite>Conor</cite>
      <time>4:17</time>
      <p>
          That   actually,   while   I   was   saying   it,   that   popped   into   my   head,   and   I   think   you&#39;re   definitely   correct   about   that.   Um   also   not   what   I&#39;m   thinking   of.   This   is   a   fundamental   property   of   what   the   algorithm   is.
      </p>

        <cite>Bryce</cite>
      <time>4:29</time>
      <p>
          Something   like   the   not   allowed   to   Oh,   it&#39;s   uh   uh   it   it   it&#39;s   it&#39;s   in   place.   It   it   modifies.
      </p>

        <cite>Conor</cite>
      <time>4:35</time>
      <p>
          Correct.   They&#39;re   permuting   algorithms.   Right.   And   that&#39;s   actually   something   that   I   noticed   when   I   was   watching   the   Tyler   Weaver   video   is   the   first   thing   he   does   is   create   a   temporary   because   he&#39;s   passing   in   by   cons   ref,   and   he   creates   a   temporary   and   just   copies   it   for   a   lot   of   them.   And   for   an   algorithm   where   you&#39;re   just   pulling   out   the   top   two   elements,   that&#39;s   kind   of   irritating   that   I&#39;m   gonna   have   to   permute   my   input   vector   or   in   for   input   range,   if   you   will,   because   this   is   really   like   a   reduction,   one   of   the   really   nice   things   I   like   about   uh   reductions   and   transform   reduces   is   that   they&#39;re   very   functional   in   the   sense   that   they&#39;re   not   modifying   the   input   range.   Um,   and   a   lot   of   other   algorithms,   like   anytime   you   reach   for   nth   elements,   and   there&#39;s   not   actually   an   equivalent   nth   element   copy,   which   a   lot   of   the   times   you   wouldn&#39;t   actually   want   this,   at   least   in   this   case,   you   wouldn&#39;t   want   that   second,   as   you   always   point   out,   Ben,   that   is   it   partition   copy   is   one   of   the   only   uh   algorithms   that   has   two   output   iterators.
      </p>

        <cite>Ben Deane</cite>
      <time>5:26</time>
      <p>
          It&#39;s   the   only   algorithm   in   the   standard   that   has   two.   And   and   actually   we   should   talk   more   about   that   in   a   minute,   because   I   think   it&#39;s   overlooked.
      </p>

        <cite>Conor</cite>
      <time>5:32</time>
      <p>
          Yeah,   and   so   and   if   you   had   an   nth   element   copy,   which   if   there&#39;s   a   partition   copy,   it   seems   like   there   should   be   an   nth   element   copy,   but   there&#39;s   a   lot   of   those   missing   sort   of   prefix-suffix   algorithms.   You   could   use   nth   element   copy   to   not   modify   the   input   sequence,   but   then   you&#39;re   gonna   end   up   with   a   range   of   n   minus   two   elements   in   you   know   that   one   of   your   iterators   pointing   to   that   you   don&#39;t   need.   And   I   think   that&#39;s   actually   I   can   screen   share   here   in   a   bit   if   you   want   to   look   at   um   the   profiling   that   uh   Tyler   did   is   nth   element   is   actually   quite   a   bit   slower   than   so   like   I   ended   up   writing   my   own   versions   and   transform,   reduce,   and   accumulate   were   definitely   the   fastest,   but   um   nth   element   was   slower   than   the   uh   max   element   and   then   erasing   a   single   value   um   and   then   doing   another   max   element.   And   I   think   that   probably   has   to   do   with   the   fact   that   you&#39;re   permuting   um   your   your   input   range.   Anyways,   thoughts.
      </p>

        <cite>Bryce</cite>
      <time>6:20</time>
      <p>
          Well,   what   one   of   the   nice   things   about   uh   the   non-destructive   algorithms,   um,   in   particular   in   the   parallel   case,   is   uh   you   can   do   multiple   different   things   in   parallel   on   the   same   data   set.
      </p>

        <cite>Conor</cite>
      <time>6:33</time>
      <p>
          Multiple   things   in   parallel.
      </p>

        <cite>Ben Deane</cite>
      <time>6:36</time>
      <p>
          If   you   have   a   read-only   data   set,   yeah.
      </p>

        <cite>Conor</cite>
      <time>6:40</time>
      <p>
          That   is   true,   yeah.   I   guess   I   don&#39;t   really   think   I   always   think   like   sort   of   data   flow   pipeline-y,   but   if   you   don&#39;t   think   that   way   and   you   just   use   it   as   sort   of   like   an   immutable   source,   then   yeah,   you   can   definitely   Well   I   mean   put   pipelines   are   just   boring   graphs,   Connor.
      </p>

        <cite>Bryce</cite>
      <time>6:58</time>
      <p>
          Just   very   boring   graphs.
      </p>

        <cite>Conor</cite>
      <time>6:59</time>
      <p>
          You   just   need   to   think   about   more   interesting   graphs   than   not   when   you   introduce   the   phi   combinator,   Bryce.
      </p>

        <cite>Bryce</cite>
      <time>7:06</time>
      <p>
          Somebody   has   just   completed   their   thesis.
      </p>

        <cite>Ben Deane</cite>
      <time>7:11</time>
      <p>
          There   was   one   more   thought   that   um   as   I   was   listening   to   your   last   to   the   to   this   podcast   to   your   podcast   about   the   algorithms,   and   Bryce   Bryce   mentioned,   I   had   the   thought,   and   Bryce   mentioned   it,   that   um   it&#39;s   interesting   to   think   about   a   kind   of   tricksy   way   to   solve   this   with   min-max   element.
      </p>

        <cite>Conor</cite>
      <time>7:27</time>
      <p>
          Is   that   possible?   You   had   the   same   thought.
      </p>

        <cite>Ben Deane</cite>
      <time>7:30</time>
      <p>
          It   reminded   me,   so   I   haven&#39;t   worked   through   it.   I   don&#39;t   it   is   definitely   tricky   to   do   because   it&#39;s   it   has   to   be   adaptive   in   the   comparator   or   something.   Yeah.   Um   but   it   reminded   me,   does   min-max   element   have   an   overload?   You   can   pass   a   comparator   to   it,   yeah.
      </p>

        <cite>Bryce</cite>
      <time>7:47</time>
      <p>
          Yeah,   of   course.   Really?   Yeah.   I   mean,   what?   Yeah,   yeah,   yeah.
      </p>

        <cite>Conor</cite>
      <time>7:51</time>
      <p>
          Keep   talking,   Ben.   Keep   talking.   This   is   great.   I   love   learning   stuff   live   on   ADSP.
      </p>

        <cite>Ben Deane</cite>
      <time>7:58</time>
      <p>
          Have   you   read   the   paper   A   Killer   Adversary   for   Quicksort?   It&#39;s   a   paper   by   Doug   McElroy.   He&#39;s   he   who   is   the   bash   Unix.   He&#39;s   often   cited   as   the   inventor   of   bash   pipelining.
      </p>

        <cite>Bryce</cite>
      <time>8:14</time>
      <p>
          Oh,   he   is   one   of   my   heroes   then,   because   I   love   bash.
      </p>

        <cite>Ben Deane</cite>
      <time>8:18</time>
      <p>
          Yeah,   he&#39;s   he&#39;s   a   great   guy.   Uh   um   I   was   lucky   enough   to   attend   one   of   his   talks   at   LambdaConf,   I   think   it   was   2015.   He   talked   about   Oh   wow.   He   presented   12   lines   of   Haskell   that   were   a   complete   a   complete   framework   for   manipulating   polynomials,   including   infinite   sequences.   And   I   that   was   just   beautiful.   Anyway,   um   highly   recommended   that   that   talk.   Uh   we&#39;ll   link   it   in   the   show   notes,   so   definitely   we&#39;ll   he   has   this   paper,   A   Killer   Adversary   for   Quicksort,   which   the   the   tenet   of   the   paper   is   um   you   know   we   all   know   that   Quicksort   performs   poorly   on   certain   inputs   that   can   go   quadratic,   right?   If   the   if   the   pivot   selection   is   poor.   Um   and   so   he   constructs   a   deliberately   bad   input   for   quicksort   that   is   constructed   adaptively   on   the   fly   to   bamboozle   QuickSort&#39;s   pivot   selection.   Okay   Um   and   it&#39;s   that   kind   of   adaptive   nature   that   I   mean   that   that&#39;s   what   it   made   me   think   of   that   when   I   thought   about   this   potential   min-max   element   solution   to   top   two.   Um   because   again   it   would   have   to   be   adaptive   in   the   comparator   somehow.
      </p>

        <cite>Conor</cite>
      <time>9:43</time>
      <p>
          Yeah,   and   I   I   brought   up   the   Min-Max   Element   page.   It   definitely   not   that   I   doubt   you,   I   just   had   to   confirm.   Um   because   I   think   I   said   actually   on   that   episode   that   Min-Max   Element   didn&#39;t   have   an   overload   version   that   takes   a   custom   comparator,   which   uh   I   apologize   to   both   Bryce   and   the   listener.   That   is   entirely   incorrect.
      </p>

        <cite>Bryce</cite>
      <time>10:02</time>
      <p>
          And   uh   it   will   be   considered   accepting   your   apology,   but   I   need   to.   I   I   appreciate   that.
      </p>

        <cite>Conor</cite>
      <time>10:08</time>
      <p>
          That&#39;s   accepted   right   there.   Yeah,   well,   I   mean   that&#39;s   actually   that&#39;s   pretty   big   uh   force.   He&#39;s   from   he&#39;s   a   new   he&#39;s   a   New   Yorker,   you   know.   Uh   he&#39;s   lucky   he   didn&#39;t   just   write   then   and   there   say,   you   know   what,   I&#39;m   done   with   this   podcast.
      </p>

        <cite>Bryce</cite>
      <time>10:22</time>
      <p>
          I   am   an   aggressive   megalomaniac,   so   that   is   pretty   big   of   me.
      </p>

        <cite>Conor</cite>
      <time>10:28</time>
      <p>
          I   appreciate   Bryce&#39;s   self-awareness,   you   know.   It&#39;s   just   that   uh   if   he   wasn&#39;t   self-aware,   I   just   uh   this   guy&#39;s   a   little   bit   uh,   but   you   know,   he&#39;s   he&#39;s   he   acknowledges   it,   so   it&#39;s   all   good.   It   is   it   is   important   to   own   one&#39;s   bullshit.   So   yeah,   maybe   um   I&#39;m   sort   of   curious   if   I   might   I&#39;m   I&#39;m   I   won&#39;t   be   surprised   at   this   point   because   we&#39;ve   had   we&#39;ve   basically   had   a   blog   slash   uh   one   blog   and   one   YouTube   video,   so   maybe   there&#39;s   a   listener   that&#39;s   gonna   go   in.
      </p>

        <cite>Bryce</cite>
      <time>10:59</time>
      <p>
          Because   I   suspect   I   will   find   reasons   to   object   to   them.
      </p>

        <cite>Conor</cite>
      <time>11:05</time>
      <p>
          Yeah,   let&#39;s   do   it.
      </p>

        <cite>Bryce</cite>
      <time>11:06</time>
      <p>
          And   this   will   be   this   will   be   uh   an   instructive   uh   episode   um   in   uh   how   to   uh   produce   good   benchmarks,   or   at   the   very   least,   benchmarks   that   pass   muster   with   me.
      </p>

        <cite>Conor</cite>
      <time>11:19</time>
      <p>
          Yeah,   so   maybe   this   will   I&#39;m   not   actually   sure   if   this   is   the   start   of   part   three   or   part   two,   but   definitely   this   is   a   great.   We&#39;re   talking   about   profiling   now.   Uh   let   me   share   my   screen.   And   uh   the   listener   is   actually   able   to   go,   we&#39;ll   link   this   in   the   show   notes   and   um   check   out.   Oh   man.   Actually,   can   you   guys   are   seeing   Twitter?   Um   at   the   top   of   whatever   part   of   the   Audacity.   Oh,   what   the   heck? Stop   sharing.
      </p>

        <cite>Bryce</cite>
      <time>11:52</time>
      <p>
          Audacity   is   the   name   of   the   software   that   we   use   to   record   the   uh   episodes.
      </p>

        <cite>Conor</cite>
      <time>11:59</time>
      <p>
          Why   does   it   keep   on   giving   you   we&#39;re   seeing   more   audacity?   Oh   yeah.   That   makes   sense.
      </p>

        <cite>Bryce</cite>
      <time>12:06</time>
      <p>
          Just   share   your   full   screen.   Share   you.
      </p>

        <cite>Conor</cite>
      <time>12:08</time>
      <p>
          I&#39;ve   got   two   m   I&#39;ve   got   two   monitors,   so   that   would   be.   That   would   be   I   am   sharing   my   full   screen,   but   I   just   can&#39;t   share   both   monitors.
      </p>

        <cite>Bryce</cite>
      <time>12:14</time>
      <p>
          Wait,   wait,   wait,   wait.   That   means   we&#39;ve   got   to   look   for   anything   embarrassing   in   the   tabs.
      </p>

        <cite>Conor</cite>
      <time>12:18</time>
      <p>
          No,   that&#39;s   not   gonna   happen.   Although,   point   free.   Should   we   I   mean   I   didn&#39;t   delete   it.   We&#39;ll   actually   take   before   we   get   to   the   profiling,   let&#39;s   take   a   we&#39;re   gonna   take   a   small   interlude   here,   Ben.
      </p>

        <cite>Ben Deane</cite>
      <time>12:32</time>
      <p>
          I   was   gonna   suggest   also   put   on   the   stack   the   algorithm   you   talked   about   from   the   meetup   last   night.   Forget   it.
      </p>

        <cite>Conor</cite>
      <time>12:40</time>
      <p>
          Was   the   uh   oh   yeah,   yeah,   the   um   yeah,   the   what&#39;s   closest   to   zero   and   choose   the   largest   of   the   two.   Alright,   so   we   got   three   things   on   the   stack.   What   we&#39;re   gonna   talk   about   for   a   second   here   is   pointfree.io   and   a   future   YouTube   video   because   I   got   s   I   got   so   excited   last   night   and   then   derailed   my   meetup   for   like   a   good   I   said   I   wasn&#39;t   gonna   talk   about   it   for   20   minutes,   and   then   I   proceeded   to   talk   about   it   for   15   minutes   and   had   to   like   literally   cut   myself   off   because   I   was   I   was   so   excited   about   it.   This   is   what   makes   the   meetup   great.   I   mean,   I   felt   it   I   was   it   was   a   little   too   I   really   sort   of   I   it   I   felt   like   it   derailed   a   little   bit   too   quickly   because   I&#39;m   not   sure   how   many   people   were   following   along,   but   I   was   so   I   was   so   excited   about   it.
      </p>

        <cite>Bryce</cite>
      <time>13:21</time>
      <p>
          People   don&#39;t   always   follow   where   you&#39;re   going,   but   they&#39;re   excited   to   to   to   be   on   the   journey   because   of   how   excited   you   are.
      </p>

        <cite>Conor</cite>
      <time>13:30</time>
      <p>
          Yeah.   So   all   right,   on   the   stack,   point   free   is   next.   Then   we&#39;re   gonna   look   at   some   profiling   uh   from   uh   Tyler,   and   then   we&#39;re   gonna   talk   about   an   algorithm   we   discussed   last   night.
      </p>

        <cite>Bryce</cite>
      <time>13:38</time>
      <p>
          So   I&#39;m   confused   about   your   stack   because   you   seem   to   have   you   seem   to   have   inserted   at   both   ends   of   the   stack.
      </p>

        <cite>Conor</cite>
      <time>13:46</time>
      <p>
          This   is   true.   All   right,   Q,   whatever.   It&#39;s   a   stack   because   we&#39;re   gonna   pop   off   in   one   direction.   But   that&#39;s   not   what   makes   it   a   heap.   It&#39;s   a   heap.   We   had   a   case.   Things   have   been   pushed   on   by   priority   and   they&#39;re   gonna   be   popped   off   in   the   other.   Exactly.   So   for   context,   I   have   a   I   have   a   programming   virtual   meetup.   Um   Ben   attends   from   time   to   time,   and   I   think   you&#39;ve   been   there   for   most   of   the   uh   seven   languages   in   seven   weeks,   and   we   were   covering   chapter   five   and   programming   language   four   of   seven   last   night,   which   was   Scala.   And   one   of   the   problems   in   the   chapter   was   to   write   a   little   tic-tac-toe   status   function   that,   given   a   board   that   is   partially   filled   out,   um,   determine   if   there&#39;s   a   winner   uh   and   which   the   if   it&#39;s   an   X   or   an   O,   or   if   there&#39;s   no   winner.   And   uh,   without   going   into   too   many   details,   at   one   point   there&#39;s   a   function   that   I   used   in   my   solution   for   both   a   Scala   solution   and   a   Haskell   solution   of   a   function   called   all   equal   to   that   takes   a   character   and   a   string.   And   it   basically   just   determines   is   the   string,   uh,   are   all   the   characters   in   that   string   equal   to   the   character   that   you   specified.   So   if   you&#39;re   given   all   equal   to   the   character   x   and   xxx,   it&#39;ll   return   true.   And   if   it&#39;s   xox   or   x   dot   x,   it&#39;ll   return   false.   And   in   Haskell,   I   had   programmed   that   by   basically   all   equal   to   uh   with   a   single   parameter   e   for   the   character,   and   then   equals   and   dot   map   parentheses   equal   equal   e   and   parentheses.   Should   we   be   spelling   this   out   for   the   listeners?   Great   radio.   But   yeah,   great   radio.   And   you   don&#39;t   really   need   to   follow   it   or   like   be   spelling   it   out   in   your   head,   but   we   went   to   this   site,   pointfree.io,   because   I   was   like,   I   could   I   I   said   just   sort   of   off   the   cuff   during   the   um   meetup,   you   could   make   this   point   free,   but   it   probably   wouldn&#39;t   um   probably   wouldn&#39;t   be   as   readable   because   right   basically   what   you&#39;re   doing   is   you&#39;re   mapping   a   unary   function   over   your   string   and   saying   and   converting   that   to   a   string   of   Booleans.   That&#39;s   uh   true   if   it&#39;s   equal   to   your   specified   character   and   false   if   not.   And   then   once   you   have   that   uh   list   of   Booleans,   you   just   do   a   logical   and   reduction.   And   if   they&#39;re   all   equal   to   true,   you   get   a   true   back.   If   any   one   of   them   is   equal   to   false,   you   get   false   back.   And   so   that&#39;s   basically   you   could   do   this   with   um   a   transform   reduce   in   C,   where   you&#39;re   transforming   your   characters   in   your   string   to   Booleans   and   then   doing   the   logical   and   reduction.   So   we   plug   this   into   point   free.io,   and   what   we   get   back,   it   changes   the   and,   which   is   our   logical   and   reduction,   to   all,   which   is   the   equivalent   equivalent   of   std   all   of   in   C.   And   basically   takes   a   list   of   Booleans,   all   takes   a   list   of   anything   and   a   unary   predicate   that&#39;s   gonna   evaluate   that   uh   list   of   things   and   turn   it   into   a   list   of   Booleans.   And   so   it   turns   AND   into   all,   keeps   the   dot,   which   is   the   B   combinator,   the   blue   bird,   everybody   loves   the   blue   bird,   and   uh   and   then   drops   the   map   and   the   you   and   it   turns   the   unary   operation   into   a   binary   operation.   And   I&#39;m   not   even   gonna   really   try   to   begin   to   explain   what   happened   there,   but   basically   the   B   combinator,   it   it   oh,   it&#39;s   I   mean,   do   I   do   I   even   attempt,   Ben?   Do   you   want   to   try?   It&#39;s   it&#39;s   just   I&#39;ll   make   a   YouTube   video   and   maybe   people   will   understand.   But   I   was   so   excited.   Jean   Michel,   who&#39;s   an   attendee,   he   pointed   out   I   was   like,   oh,   I   think   I   understand   what&#39;s   going   on.   I   didn&#39;t   even   notice   at   first   that   the   and   changed   to   an   all.   This   website   is   pure   magic.   Uh   Ben,   do   you   have   anything   to   add?
      </p>

        <cite>Ben Deane</cite>
      <time>17:18</time>
      <p>
          Uh   I   would   just   say   for   the   listener,   the   point   of   the   website   is   it&#39;s   called   point   free   because   it   it   changes   functions   to   the   point   of   the   function   is   the   argument   of   the   function   or   the   parameter   to   the   function,   right?   And   so   it   it   removes   function   parameters   in   favor   of   plain   function   composition,   hence   point   free.
      </p>

        <cite>Bryce</cite>
      <time>17:37</time>
      <p>
          So   in   in   instead,   like   the   C   analogy,   instead   of   having   a   you   know,   sequence   of   calls   to   algorithms   where   the   inputs   are   explicitly   named,   having   a   pipeline   of   uh   range   adapters   that   can   be   applied   to   any   input   where   the   the   uh   the   input   sequence   is   an   implicit   in   past   between   all   of   them.   Or   you   know,   the   other   analogy   is   like   you   know,   a   bash   pipeline,   um,   where   you   don&#39;t   explicitly   name   uh   uh   you   know   stood   in.   It&#39;s   just   implicitly   passed   from   thing   to   thing.
      </p>

        <cite>Conor</cite>
      <time>18:14</time>
      <p>
          Yeah,   Eric   Niebler   was   the   first   one   that   pointed   because   I   think   at   some   point   I   remarked   in   a   meeting   or   in   a   conversation   he   was   a   part   of   that   I   said,   oh,   C   doesn&#39;t   have   support   for   point-free.   And   he   said,   Ah,   that&#39;s   not   actually   true.   If   you   are   building   up   a   custom   range   adapter,   that   is   the   composition   of   you   know   a   transform   and   a   filter   and   something   else,   um,   that&#39;s   technically   point-free,   because   you   never   mention   your   arguments,   um,   which   I   was   like,   oh   yeah,   that&#39;s   a   good   point.
      </p>

        <cite>Bryce</cite>
      <time>18:36</time>
      <p>
          Um   at   the   very   least,   you   you   never   mention   your   primary   argument.   In   a   lot   of   cases   there,   you   might   have   an   argument   of   a   functor   or   or   some   other   parameter   to   one   of   those   functions,   but   but   not   the   actual   input   sequence.
      </p>

        <cite>Conor</cite>
      <time>18:50</time>
      <p>
          Yeah,   not   not   the   argument   that   will   be   required   for   basically   the   the   callable   that   is   returned.   Um   yeah,   and   maybe   well,   we&#39;re   probably   not   gonna   have   time   to   get   into   all   these   things,   but   yeah.   I&#39;ve   been   thinking   a   lot   about   higher   order   function   these   functions   these   days.
      </p>

        <cite>Ben Deane</cite>
      <time>19:06</time>
      <p>
          Well   I   was   gonna   say   you   made   the   observation   here,   Conor.   So   so   the   dot   operator   is   is   the   bluebird   combinator,   right?   Um   function   composition.   Yep.   Uh   and   you   you   come   at   this   code   very   much   in   a   combinator   mindset   where   at   least   you   said   last   night   you   were   you   think   about   combinators,   and   the   way   you   think   about   combinators,   you   know,   you   can   tell   me   I&#39;m   wrong,   but   um   you   think   of   them   as   functions   of   one   argument.   Well,   in   particular,   the   the   I   guess   the   bluebird   combinator   combines   together   two   functions   of   one   argument.   Right.   And   looking   at   this   Haskell   point-free   version   uh   is   not   obvious   from   that   mindset,   because   of   course   the   equality   operator   is   a   binary   operator,   it&#39;s   not   a   function   of   one   argument,   it&#39;s   a   function   of   two   arguments.   Uh   and   indeed   all   is   a   function   of   two   arguments   as   well.   And   so   uh   it   it   at   first   glance,   you   know,   in   that   mindset,   it   looks   like   how   is   how   is   dot   working   to   combine   these   two   functions   because   they&#39;re   not   functions   of   one   argument,   uh   except   of   course   that   in   Haskell   every   function   is   a   function   of   one   argument,   which   can   then   return   a   function   as   opposed   to   returning   the   eventual   result   because   of   partial   application   being   built   into   the   language.
      </p>

        <cite>Conor</cite>
      <time>20:17</time>
      <p>
          Yep.   That&#39;s   and   that&#39;s   that&#39;s   like   a   very   even   to   this   day,   I   have   to   like   when   I   start   thinking   about   this   stuff,   I   like   I   have   to   sit   down   and   start.   I   think   like   at   some   point   I&#39;m   gonna   have   to   go   and   implement   a   mini,   it   doesn&#39;t   have   to   be   Haskell,   but   basically   a   language   that   has   currying   built   in   because   it   is   it&#39;s   a   very   different   model,   like   you   said,   right?   In   this   code,   all   dot   you   know,   paren   equal   equal   paren.   That&#39;s   uh   the   B   combinator   which   composes   two   unary   functions   sitting   between.   So   it&#39;s   an   infix   operator.   It   sits   between   two   by   two   binary   functions   that   so   two   functions   that   take   two   arguments.   Right.   Somehow,   and   somehow   it   works.   And   somehow   it   works.   It&#39;s   very   even   like   the   uh   the   blackbird,   the   b1   combinator,   one   of   my   favorite   combinators,   we   talked   about   it   in   a   previous   episode,   where   that   that   combines   a   unary   function   and   a   binary   function.   There&#39;s   a   definition   of   that   where   basically   three   beat   combinators   in   a   row,   three   bluebirds   in   a   row,   equals   the   blackbird.   And   like   I   have   no   idea,   like   I   have   not   sit   down   to   like   work   through   the   eta   reduction   or   whatever   you   whatever   technique   you   use.   But   like   I&#39;ve   seen   that   in   a   talk   and   a   couple   other   blogs,   and   like   I&#39;m   just   like,   I   don&#39;t   even   like   BBB,   I   do   not   understand   how   that   equals   B1,   but   like   that&#39;s   the   whole   I   mean,   Haskell   Curry&#39;s   very   first   paper   on   the   topic   in   1929   was   entitled   An   Analysis   of   Logical   Substitution,   and   before   he   had   named   it   combinatory   logic,   like   that&#39;s   how   he   approached   it.   Like   programming   languages   weren&#39;t   even   a   thing   back   then,   pre   you   know,   30s   or   40s,   and   it   was   it   was   substituting,   you   know,   one   expression   for   another   expression   based   on   these   sort   of   relationships.   And   uh   yeah,   I   I&#39;m   it   I   you   are   definitely   correct   that   I   come   at   it   from   like   a   very   specific,   you   know,   mindset   that&#39;s   primarily   of   like   how   they&#39;re   used   in   APL,   and   then   I&#39;ve   been   sort   of   Reverse   engineering   my   way   into   these   other   sets,   you   know,   or   or   contexts   of   Haskell   and   I   came   to   uh   Haskell   first   before   combinators.
      </p>

        <cite>Ben Deane</cite>
      <time>22:11</time>
      <p>
          So   I   think   about   this   just   in   terms   of   how   the   types   unify   and   check.   Pretty   much.   Interesting.
      </p>

        <cite>Conor</cite>
      <time>22:18</time>
      <p>
          And   it&#39;s   interesting   because   technically   I   I   I   landed   at   Haskell   first,   but   didn&#39;t   really   appreciate   what   the   dot   did   or   really   what   the   dollar   sign   did.   I   just   sort   of   knew   that   if   I   stuck   them   in   different   places,   oh   then   it&#39;d   work.   And   um   and   so   yeah,   even   though   I   saw   them   first   in   Haskell,   I   didn&#39;t   really   appreciate   what   they   were   until   coming   back   from   from   array   languages.
      </p>

        <cite>Ben Deane</cite>
      <time>22:37</time>
      <p>
          It&#39;s   well,   it&#39;s   spring,   it&#39;s   a   good   time   to   study   combinators.   I   would   say   I   currently   have   two   combinators   living   outside   my   front   door.   I   have   there&#39;s   a   I   two   finches   made   a   nest   in   my   wreath.   And   for   the   last   month   I   have   been   using   my   front   door   as   a   result.   They   they   laid   she   laid   five   eggs,   of   which   four   hatched,   and   there   are   currently   at   least   two   baby   birds   that   are   still   surviving.   Oh,   but   you   know,   it&#39;s   it&#39;s   sad   that   some   die,   but   that&#39;s   nature   for   you.   That&#39;s   why   that&#39;s   why   she   had   that   number,   I   suppose.   But   yes,   they&#39;re   they&#39;re   surviving   and   thriving,   as   far   as   I   can   tell.
      </p>

        <cite>Conor</cite>
      <time>23:20</time>
      <p>
          And   the   for   for   the   F   combinator   is   definitely   uh   is   a   combinator.
      </p>

        <cite>Bryce</cite>
      <time>23:25</time>
      <p>
          Then   we&#39;re   gonna   we&#39;re   gonna   need   some   pictures.
      </p>

        <cite>Conor</cite>
      <time>23:28</time>
      <p>
          Yeah.   To   add   to   our   show   notes.   And   uh,   I   do   have   a   picture.
      </p>

        <cite>Ben Deane</cite>
      <time>23:33</time>
      <p>
          Yeah,   uh,   we   don&#39;t   disturb   them   very   much,   but   I   do   have   one   picture.
      </p>

        <cite>Bryce</cite>
      <time>23:38</time>
      <p>
          I   assumed   that   you   had   some   some   pictures.
      </p>

        <cite>Conor</cite>
      <time>23:42</time>
      <p>
          Yeah,   for   and   for   the   listener   that   hasn&#39;t   heard   the   lore   of   Raymond   Smullian&#39;s   uh   to   mock   a   mocking   bird   and   why,   so   a   mathematician,   famous   mathematician   that   um   I   think   I   probably   heard   about   it   from   you,   Ben,   um,   or   maybe   if   I   first   heard   about   uh   the   book   from   that   uh   Point   Free   or   Die   Talk   that   was   given   by   Amar   Shah   back   in   like   Strange   Loop   2016.   But   then   you   mentioned   um   Raymond   Smullian   and   and   had   told   me   that   he   was   a   very   well-known   and   prolific   writer   of   sort   of   logic   puzzle   books   and   math   puzzle   books.   And   he   wrote   this   book   in   1985   called   Tamaka   Mockingbird,   where   he   nicknamed   all   the   combinators.   And   the   reason   he   gave   them   bird   nicknames   is   because   very   famously   or   not   so   famously,   because   I   only   learned   this   from   reading   um   Tamaka   Mockingbird,   is   that   Haskell   Curry,   who   was,   you   know,   basically   did   all   of   the   research   that   led   to   the   field   of   combinatory   logic,   he   was   a   huge   avid   bird   watcher,   and   that&#39;s   why   it   was   basically   an   homage   to   him.   And   and   the   funny   thing   is,   is   I   was   at   lunch   the   other   day   and   I&#39;m   pretty   sure   I   saw   a   warbler   because   I   actually,   through   like   investigating   combinators   and   spending   so   much   time   researching   them   and   tweeting   out   photos,   I   actually   am   starting   to   uh   recognize   types   of   birds.   And   I   like   I   fell   off   my   chair   and   like   ran   outside   the   restaurant   to   try   and   get   a   photo   of   this   warbler   because   I   was   like,   a   warbler   in   the   wild.   That&#39;s   the   W   combinator,   it&#39;s   one   of   the   best   ones.
      </p>

        <cite>Bryce</cite>
      <time>25:01</time>
      <p>
          Connor&#39;s   gonna   become   a   bird   watcher.   Connor&#39;s   gonna   become   like   a   bird   man.   He&#39;s   gonna   have   like   a   collection   of   birds   that   he   cares   for.
      </p>

        <cite>Ben Deane</cite>
      <time>25:11</time>
      <p>
          This   week,   what   may   be   useful,   I   just   heard   about   this   app   that&#39;s   out   of   um   Cornell.   Merlin.   Merlin?   Merlin   Bird   ID?   Yeah.
      </p>

        <cite>Conor</cite>
      <time>25:24</time>
      <p>
          Yeah,   yeah,   it&#39;s   like   the   you   just   record   some   audio   and   then   it   tells   you   what   it   is.   But   apparently   they&#39;re   doing   amazing   things   identifying   birds.
      </p>

        <cite>Bryce</cite>
      <time>25:33</time>
      <p>
          About   about   10   years   ago,   my   dad,   who&#39;s   a   bird   watcher,   like   had   that   idea   like   for   an   app,   and   he   was   like,   Can   this   be   made?   And   I&#39;m   like,   I&#39;m   sure   it   can   be,   but   I&#39;m   not   the   person   to   make   it.
      </p>

        <cite>Conor</cite>
      <time>25:50</time>
      <p>
          Alright,   moving   on   to,   unless   do   we   have   anything   we   want   to   say   about   point-free   tacit   programming   before   we   move   on   to   uh   the   profiling?
      </p>

        <cite>Bryce</cite>
      <time>25:57</time>
      <p>
          We   should   go   on   to   the   profiling   because   Bryce   is   uh   I&#39;m   fading   fast,   it&#39;s   approaching   my   bedtime.   I   don&#39;t   know   how   you&#39;re   still   alive   given   that   you   exercised   more   than   me   today.
      </p>

        <cite>Conor</cite>
      <time>26:09</time>
      <p>
          Uh   I&#39;m   wired.   I&#39;m   wired.   I   mean   uh   we&#39;re   talking   about   birds   combinators.
      </p>

        <cite>Bryce</cite>
      <time>26:13</time>
      <p>
          Yeah,   the   crash   hasn&#39;t   happened   yet.   It&#39;s   coming.   It&#39;s   coming.
      </p>

        <cite>Conor</cite>
      <time>26:16</time>
      <p>
          I   mean   you   also   have   a   way   earlier   bedtime   than   me,   you   know.   I   typically   go   to   sleep   at   like   midnight.
      </p>

        <cite>Bryce</cite>
      <time>26:19</time>
      <p>
          No,   my   my   bedtime   has   been   disrupted   the   entire   winter.   Not   entirely   because   of   winter,   but   because   of   my   bring   up   the   profiling   make   Bryce   nice   and   angry   so   he   stays   awake.
      </p>

        <cite>Ben Deane</cite>
      <time>26:32</time>
      <p>
          There   are   things   wrong   here,   Bryce.
      </p>

        <cite>Conor</cite>
      <time>26:35</time>
      <p>
          Thanks   for   listening.   We   hope   you   enjoyed   and   have   a   great   day.
      </p>