Fix issue with memory allocations in lisp::serialise

Was to do with sv_make.

lib/sv.c

@@ -60,7 +60,8 @@ sv_t sv_chop(sv_t sv, u64 size)
 
 sv_t sv_concat(arena_t *allocator, sv_t a, sv_t b)
 {
-  sv_t c = sv_make(allocator, a.data, a.size + b.size);
+  sv_t c = sv_make(allocator, NULL, a.size + b.size);
+  memcpy(c.data, a.data, a.size);
   memcpy(c.data + a.size, b.data, b.size);
   return c;
 }

lisp/lisp.c

@@ -252,7 +252,13 @@ sv_t serialise(context_t *ctx, lisp_t *ptr)
         lmember = CAR(lmember);
 
       sv_t member = serialise(ctx, lmember);
-      s           = sv_concat(&ctx->scratch, s, member);
+      if (!s.data)
+        s = member;
+      else
+        s = sv_concat(&ctx->scratch, s, member);
+
+      // NOTE: eventually this separator we add per item should be mutable at
+      // runtime.
       if (IS_TAG(lsp, CONS) && !CDR(lsp))
         continue;
       else if (IS_TAG(CDR(lsp), CONS))

oats.org

@@ -4,15 +4,6 @@
 #+FILETAGS: :oats:
 
 * Issues :issues:
-** TODO Fix issue with memcpy overlap when string concatenating
-[[file:lisp/lisp.c::// FIXME: Something is going wrong here!]]
-
-Ideas on what's going wrong:
-- String sizes seem off
-- Maybe something is wrong with arena allocator; we use
-  [[file:lib/sv.c::newsv.data = arena_realloc(allocator, sv.data,
-  sv.size, newsv.size);][arena_realloc]] which seems to be the root of
-  the memcpy-overlap
 * Features :features:
 ** WIP Reader :reader:
 We want something a bit generic: able to handle reading from some
@@ -205,3 +196,18 @@ more work than the lowest byte (to look at it we'll need to push that
 byte all the way down).  So we'll be going for a low byte strategy by
 shifting the pointer up by one byte.  This leaves us with the lowest
 byte to play with as we choose.
+** DONE Fix issue with memcpy overlap when string concatenating
+[[file:lisp/lisp.c::// FIXME: Something is going wrong here!]]
+
+Ideas on what's going wrong:
+- String sizes seem off
+- Maybe something is wrong with arena allocator; we use
+  [[file:lib/sv.c::newsv.data = arena_realloc(allocator, sv.data,
+  sv.size, newsv.size);][arena_realloc]] which seems to be the root of
+  the memcpy-overlap
+
+2025-05-30 08:11:24: Was an issue with ~sv_make~; if you pass a data
+pointer, it will try to copy the size you passed as bytes from the
+pointer.  In ~sv_concat~ we pass the overall new size of the string
+we're allocating and pass in the first argument to copy in the initial
+part of the data first.