Files
ioccc/todo.md
T
Cody Boone Ferguson cfb9c66fae Various updates to todo.md
What prompted this update is a 'problem' I noticed in 2019/duble: it
leaves sockets lying round in the current working directory but it might
not be safe for make clobber to delete them: involves '.' files and we
don't know if the glob will be different in different systems. In macOS
it appears to be '\.[A-Z]+*'
2023-05-31 02:53:46 -07:00

5.3 KiB

A todo list of known things to check and/or do

Last updated: Wed 31 May 2023 09:52:55 UTC

More things will likely be added over time and as items are done they should be removed from the list IF AND ONLY IF (!!) we are absolutely certain it is done (many times it's thought that a task was done only to find out it wasn't). It is hoped that this will be referenced and updated as this is an easy way to keep track of things that have been done already and what hasn't been done (in this file the case is mostly NOT been done).

  • Find entries with bugs as well INABIAF and add to bugs.md. For instance a INABIAF section should be added to bugs.md for 2004/jdalbec based on the README.md file. As far as I am aware the ones that have not been checked are > 2011 (and two entries in 2011, vik and zucker) and possibly some of the earlier years. I (Cody Boone Ferguson) have spent too much time when working on formatting the README.md files on finding problems out (and in some cases fixing or trying but not yet succeeding to fix an entry); if I did not do that I would be done with fixing the README.md files (format wise and mostly typos too). Of course when working on finding bugs it might necessitate updating the README.md file but that can be done at that time. I might or might not end up following this idea but I probably should so this needs to be noted.

  • Resolve as many of the issues in bugs.md with the exception of the INABIAF ones. This is likely not possible for all (one in particular comes to mind that abuses a bug in a now very old gcc version - this obviously cannot be fixed) but many certainly can be (and have been) fixed.

  • Check for typos in README.md files. Most years have been done. I (Cody) need to check the late 80s through mid 1990s plus the last two entries in 2011 and the following years. At the same time the formatting is being fixed.

  • Go through all year README.md files that is 1984/README.md, 1985/README.md and so on. As of Sun 30 Apr 2023 13:45:52 UTC the following years have been finished: 1984, 1985, 1986, 2004, 2005.

  • Verify which README.md files have not been checked for correct formatting and fix any issues.

    • This includes author information: not only at first column but making sure that the lines end with spaces so that the rendered output does not end up on one line only.
    • Cody Boone Ferguson has done many and has some kind of record of what he's done but these can be added later by having a list of years to check.
    • NOTE: as of 03 May 2023 years up through most of 2011 have been checked except that a few years I am pretty sure I did not check typos (I'd say about 1987 or 1988 through 1995 should be rechecked for typos). It's very likely I won't keep this updated until it's finished but I wanted to note it now.
    • NOTE: a while back (as of 31 May 2023) Cody noticed that some of the files he already checked could be improved wrt formatting so he'll have to have a quick look at the previously edited files; some he has done already but the vast majority have not been checked. These should be relatively easy to fix but will only be done after the rest of the files are processed (presumably with complete changes made :-) ).
  • Remove addresses from older winning entries but (if known) keep country code + add country code to winners.csv. As of half past 4 Pacific time on 22 April 2023 only the years through 1986 have been done entirely. This should not take much time but I (Cody) am bored with it for now and would rather do other more important things. This will likely not be updated until after it's done (and when I remember to remove it - I expect that when I go for it I'll do it in one sitting or maybe a few sittings with a few brief breaks).

  • Verify that 2004/gavin's img/fs.tar contains the same version of files as under 2004/gavin itself. This should be verifiable by extracting the tarball in the entry's img/ subdirectory and then typing the below from the entry's main directory (2004/gavin), making sure no output is displayed:

    diff -r . img/fs|grep -v 'Only in'

  • Try and see if I (Cody) can make 1985/sicherman even more like its original form. I have some ideas here but it might involve some discussion too. See 22 May 2023 commits for the improvements already made.

  • Check all Makefiles for notes / warnings on compilation and see if they can be moved to after the compilation as they're more likely to be seen this way. In some cases this will have to be done in the Makefile via || : (that might be good to have in the Makefiles anyway so that additional rules / targets are run) or else they'll have to be put above the compilation.

    • Make compilation of targets use || : in the Makefiles to allow additional targets to compile and/or let warnings/notes be printed. An important use of this is that some entries won't compile with clang due to defects but the alternate version will. Whether this is a good idea is TBD but it's noted here as a reminder to be discussed.