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