Skip to content
Fetching contributors…
Cannot retrieve contributors at this time
90 lines (63 sloc) 4.06 KB
TODO List for File::Util
Not necessarily listed in order of priority:
1) [DONE] Set up formal (not just private) GIT repository
2) [DONE] Separate documenation examples into a cookbook (POD)
3) [DONE] Gradually transform methods to accept input parameters in a
*::Class-like style (hashrefs, etc) while preserving backward compatibility
4) [DONE] Unicode support for reading/writing files, which just hasn't been
requested but is now there for completeness. This item has become an all-out
quest to introduce full unicode support for both file/directory names and
file encodings (on platforms where it is supported), but given the great
minefield of problems with unicode on windows regarding file names, and
directories, that may never become a reality until Perl itself "fixes" the
problem.
5) [DONE] Transform (where useful) various methods to accept callbacks.
File::Util::list_dir() is the primary target
6) [DONE] Set up File::Util::list_dir() to take a listref of regexes
7) [DONE] Continue improving the distribution so as to make it compliant
with Fedora packaging standards; it is already a maintained package for
Ubuntu and ActiveState.
8) [DONE] Code clean up in POD documentation examples.
9) [DONE] Improve and simplify code examples in POD documentation.
10) [IN PROGRESS] Constantly improve test suite until Devel::Cover
scores are something to be proud of. We're doing much better
than we used to only less than 1 month ago
11) [DONE] Remove all traces of old invocation syntax from the POD
12) [DONE] Now that documentation has been divided into the ::Manual and
the ::Cookbook, I want to add more content to them. More examples,
More fully-functional programs (recipes).
13) [DONE] Create a less-verbose Exception class (right now we only have
Exception::Diagnostic, which is overkill for power users and "experts" who
just want a quick, clean error message instead of a full-page printout of
what went wrong, how to fix it, and whose fault it was -- followed by a
callstack. We need the less verbose alternative now.
14) [DONE WITH CAVEATS*] Create a benchmark suite, plot and publish metrics as
part of the documentation.
*Turns out it wasn't a completely good idea, given that benchmarking is of
little use without context, and that means comparing the performance of
File::Util to other distributions out there that do similar things. This
has two problems, the first is that distributions are constantly evolving
and benchmarks recorded on one day are inaccurate the next. The second
is that some of the numbers are NOT flattering. In some test scenarios
File::Util has out-performed File::Find::** by 400%.
Publishing things that are potentially embarrassing is rude and I won't
do it in the tone of tooting my own horn. Mentions made to the increased
performance of File::Util however are not out-of-bounds and so will be
included in the formal documentation in defense of any claims that it is
"slow" due to its perceived size and file modularity.
One of the best things to come out of this was that I was able to increase
the performance of list_dir() by about 400% with the help of Devel::NYTProf
and also identify other bugs that I wouldn't have otherwise discovered.
15) [DONE] Memo-ize pattern "gathering" for recursive list_dir() calls for
greater efficiency. Right now there's a gather op for ever recursion, and
that just isn't necessary. That can be optimized out.
16) Provide an option to follow symlinks in list_dir(),
17) [DONE] keep track of inodes seen while traversing directories to avoid
filesystem loops
18) [DONE] Write even more unicode tests
19) Create option to allow user to specify that atomic file operations should
be performed instead of regular IO. It's a handy feature that will be
added in soon
20) [DONE] Allow users to specify custom IO layers for reads/writes/opens
21) Revisit documentation structure to see if improvements can't be made
22) Create convenience methods for UTF8 reads/writes/opens
Something went wrong with that request. Please try again.