todo.txt

Last

Last2

answer Ben about his markdown idea? meh

restore some elements from the deleted Makefile in

https://github.com/aryx/syncweb/commit/8cc3f10599bf65ab57947950f30cb6b18d83db14

once graph_code, codegraph, the parser_xxx are added to OPAM,

can start to also package the indexer and lpizer

last1

see autodefs.ocaml script in fork-tiger/old/

to pretty print better ocaml code. Try on VCS.nw? just copy what noweb does in syncweb -to_tex

fix the int x = …; that leads to

a Uses x probably because graph_code_c unsugar such decl as a decl + separate assign.

populate also hdefs_and_uses_of_chunkid

and support \swuses{} and use it for lname, etc? also for rules? meh.

index external functions, put ‘print (stdio.h)’

now that mk index use header deps (to have a low number of lookup failures), I can use those header information to improve the automatic indexing

Features

literate programming automatic index

step1: DONE get an index displayed step2: DONE get complete and correct index displayed (use codegraph? cover fields?) step3: SEMI get nice index displayed (in margin? mini indexes? no duplicate, sorted)

syncweb/noweb

handle twocolumns format, right now the chunk crossrefs triangle

are at the wrong place

have index on double columns?

\twocolumn causes a pagebreak

can use makeindex? look what I did for my PhD thesis

less: [<xxx|ttt>]

so can reference right chunk! maybe no need now that I use candidate strategy, maybe for OCaml it can be useful, but not even sure

maybe do not add subpageref for entities that are

defined partly in the same section (when functions defined by multiple pieces)

syncweb/codegraph

later: generate refs for stuff from stdlib?

nil, strlen, etc? with reference to Libcore?

later: codegraph for C via 5c-in-ocaml, so have use deps for fields too

long, lots of stuff to fix in 5c-in-ocaml …

but codegraph does not have info for fields actually, maybe better to start from 5c-in-ocaml :)

Also it adds many dependencies. Maybe easier to start from scratch and precisely from 5c-in-ocaml?

resume 5c-in-ocaml on code of mk

parse correctly everything?

fix LOC and preprocessor

need take pen and paper and go through … maybe can save result and present it in Compiler.nw or something. Hard to understand tables.

start generate the defs

start generate the uses

need resolve, but have type information so should be doable.

look at code of graph_code_clang.ml because when I was using clang

I had to dedupe files because I was also processing preprocessed ASTs

syncweb/ctwill mini indexes

later: generate mini-index a la ctwill?

or maybe better the mini-index a la noweb in the end? see below

can use ctwillmac with latex?

try hand generate some \mini reread milp.pdf? => I can generate everything from scratch, without even using noweb or cweb

but seems hard. plain tex and latex seems incompatible

send email to Hanson and Ramsey to learn about mini-index on

left and right places? Or can incorporate ctwill macros?? Or look in .aux file a la nuweb and generate myself?

later: can noweb generate mini-indexes like ctwill does?

it generates defs/uses after each chunk, nice, but I prefer the mini-indexes I think dave hanson managed to do it

build mini-indexes like in Hanson books, like ctwill of Knuth,

but filter to only put important crossref? like not the one available on the same page?

apparently according to http://www.literateprogramming.com/best.pdf the index information is generated in raw form from noweb in latex file, and david hanson wrote Icon program to dig in .aux file and build the mini-indexes

use margin notes?