| 1 | Always check out the latest perl5-porters discussions on these subjects |
| 2 | before embarking on an implementation tour. |
| 3 | |
| 4 | Bugs |
| 5 | remove recursion in regular expression engine |
| 6 | fix memory leaks during compile failures |
| 7 | make signal handling safe |
| 8 | |
| 9 | Tie Modules |
| 10 | VecArray Implement array using vec() |
| 11 | SubstrArray Implement array using substr() |
| 12 | VirtualArray Implement array using a file |
| 13 | ShiftSplice Defines shift et al in terms of splice method |
| 14 | |
| 15 | Would be nice to have |
| 16 | pack "(stuff)*", "(stuff)?", "(stuff)+", "(stuff)4", ... |
| 17 | contiguous bitfields in pack/unpack |
| 18 | lexperl |
| 19 | bundled perl preprocessor/macro facility |
| 20 | this would solve many of the syntactic nice-to-haves |
| 21 | use posix calls internally where possible |
| 22 | gettimeofday (possibly best left for a module?) |
| 23 | format BOTTOM |
| 24 | -i rename file only when successfully changed |
| 25 | all ARGV input should act like <> |
| 26 | report HANDLE [formats]. |
| 27 | support in perlmain to rerun debugger |
| 28 | regression tests using __DIE__ hook |
| 29 | lexically scoped functions: my sub foo { ... } |
| 30 | the basic concept is easy and sound, |
| 31 | the difficulties begin with self-referential |
| 32 | and mutually referential lexical subs: how to |
| 33 | declare the subs? |
| 34 | lexically scoped typeglobs? (lexical I/O handles work now) |
| 35 | wantlvalue? more generalized want()/caller()? |
| 36 | named prototypes: sub foo ($foo, @bar) { ... } ? |
| 37 | regression/sanity tests for suidperl |
| 38 | iterators/lazy evaluation/continuations/first/ |
| 39 | first_defined/short-circuiting grep/?? |
| 40 | This is a very thorny and hotly debated subject, |
| 41 | tread carefully and do your homework first |
| 42 | generalise Errno way of extracting cpp symbols and use that in |
| 43 | Errno, Fcntl, POSIX (ExtUtils::CppSymbol?) |
| 44 | the _r-problem: for all the {set,get,end}*() system database |
| 45 | calls (and a couple more: readdir, *rand*, crypt, *time, |
| 46 | tmpnam) there are in many systems the _r versions |
| 47 | to be used in re-entrant (=multithreaded) code |
| 48 | Icky things: the _r API is not standardized and |
| 49 | the _r-forms require per-thread data to store their state |
| 50 | cross-compilation support |
| 51 | host vs target: compile in the host, get the executable to |
| 52 | the target, get the possible input files to the target, |
| 53 | execute in the target (and do not assume a UNIXish shell |
| 54 | in the target! e.g. no command redirection can be assumed), |
| 55 | get possible output files back to to host. this needs to work |
| 56 | both during Configure and during the build. You cannot assume |
| 57 | shared filesystems between the host and the target (you may need |
| 58 | e.g. ftp), executing the target executable may involve e.g. rsh |
| 59 | a way to make << and >> to shift bitvectors instead of numbers |
| 60 | |
| 61 | Possible pragmas |
| 62 | debugger |
| 63 | optimize (use less qw[memory cpu]) |
| 64 | |
| 65 | Optimizations |
| 66 | constant function cache |
| 67 | switch structures |
| 68 | foreach(reverse...) |
| 69 | cache eval tree (unless lexical outer scope used (mark in &compiling?)) |
| 70 | rcatmaybe |
| 71 | shrink opcode tables via multiple implementations selected in peep |
| 72 | cache hash value? (Not a win, according to Guido) |
| 73 | optimize away @_ where possible |
| 74 | tail recursion removal |
| 75 | "one pass" global destruction |
| 76 | rewrite regexp parser for better integrated optimization |
| 77 | LRU cache of regexp: foreach $pat (@pats) { foo() if /$pat/ } |
| 78 | |
| 79 | Vague possibilities |
| 80 | ref function in list context? |
| 81 | make tr/// return histogram in list context? |
| 82 | loop control on do{} et al |
| 83 | explicit switch statements |
| 84 | built-in globbing |
| 85 | compile to real threaded code |
| 86 | structured types |
| 87 | autocroak? |
| 88 | modifiable $1 et al |