todo tweaks

raehik · Nov 29, 2023 · a3aaba2 · a3aaba2
1 parent 98e9d22
commit a3aaba2
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 27 deletions.
diff --git a/TODO.md b/TODO.md
@@ -1,17 +1,25 @@
 # binrep to-dos
-## Test new parser errors
-Where we keep track of the offset. It's very weird and probably bad :(
-
-## Intermediate types should use the bytestring builder type, not a bytestring
-Saves on allocations if you're just gonna serialize. Specifically, the text
-modules probably want this. Not sure how to do it exactly to retain maximum
-soundness.
+## Write more primitives
+  * Varint
+    * base done, but doesn't support everything
+    * `newtype Varint sign end = Varint { unVarint :: VarintRep sign }`
+    * apparently `Varint s 'LE` would be LEB128, `Varint s 'BE` would be VLQ
+    * https://developers.google.com/protocol-buffers/docs/encoding , Mason
+    * byte-oriented, so each byte is 0-127 instead of 0-255.
+    * protobuf uses little endian, twos complement
+    * Wikipedia suggests we could support signed & unsigned:
+      https://en.wikipedia.org/wiki/LEB128
+  * ZigZag
+    * Maps signed to unsigned
+    * `newtype ZigZag size end = ZigZag { unZigZag :: I 'S size end }`
+    * https://developers.google.com/protocol-buffers/docs/encoding
+    * https://hackage.haskell.org/package/zigzag-0.0.1.0/docs/Data-Word-Zigzag.html
 
-This might be a big change. I need to check aeson's design more thoroughly, it'd
-look like the `Encoding` stuff.
+## Define types for some common file formats
+* How about a shared type for xz
 
-2023-02-19: No, actually, I think this is a Text-unique thing, and they just
-need some extra definitions.
+## Test new parser errors
+Where we keep track of the offset. It's very weird and probably bad :(
 
 ## Octet instead of byte?
 Is it better to refer to octets instead of bytes? An octet is always 8 bits,
@@ -26,22 +34,6 @@ Consider implementing "practical maxes" for various types.
 
 May help prevent unexpected OOMs?
 
-## More primitives
-  * Varint
-    * base done, but doesn't support everything
-    * `newtype Varint sign end = Varint { unVarint :: VarintRep sign }`
-    * apparently `Varint s 'LE` would be LEB128, `Varint s 'BE` would be VLQ
-    * https://developers.google.com/protocol-buffers/docs/encoding , Mason
-    * byte-oriented, so each byte is 0-127 instead of 0-255.
-    * protobuf uses little endian, twos complement
-    * Wikipedia suggests we could support signed & unsigned:
-      https://en.wikipedia.org/wiki/LEB128
-  * ZigZag
-    * Maps signed to unsigned
-    * `newtype ZigZag size end = ZigZag { unZigZag :: I 'S size end }`
-    * https://developers.google.com/protocol-buffers/docs/encoding
-    * https://hackage.haskell.org/package/zigzag-0.0.1.0/docs/Data-Word-Zigzag.html
-
 ## Generate (human-readable) schema from type
 I think I do this by writing yet another typeclass, filling it out for my
 primitives, then writing yet another generic typeclass. But I'd like to make

diff --git a/TODO-haskpatch.md → haskpatch/TODO-haskpatch.md b/TODO-haskpatch.md → haskpatch/TODO-haskpatch.md