Revision history for Data-Roundtrip
0.01 2020-04-13
First version, released on an unsuspecting world.
0.02 2020-04-14
Changed NAME in documentation of the scripts, some
POD errors in lib/Data/Roundtrip.pm and some errors
in the documentation.
0.03 2020-04-15
Makefile.PL changes thanks to marto @ Perlmonks.org
0.04 2020-04-15
Added perl2dump_filtered(), perl2dump_homebrew()
and dump_perl_var_recursively().
Fixed scripts to work with new sub names.
0.05 2020-04-16
Makefile.PL minor modification.
0.07 2020-04-20
Changed default options for scripts so that they don't do pretty
printing by default.
Added import parameters 'no-unicode-escape-permanently'
and 'unicode-escape-permanently' for efficient use of *2dump*
subs.
Added benchmarks to time 2dump subs and inserted target in Makefile.
0.08 2020-04-20
Removed dependency to Sub::Override, overriding subs is done by us now.
Fixed POD mistakes.
0.09 2020-04-20
Added dependency Test::Script and Test::Pod (thank you Slaven Rezić)
0.10 2020-04-20
Added minimum version for Test::Script to be 1.09
to support script_stderr_is. Thank you Slaven Rezić!)
0.11 2020-04-21
Checked failures for test script t/02-random-data-structure.t
and provided a fix and added a warning in CAVEATS section.
0.12 2023-03-09
Wrapped within an eval{}, call to JSON::decode_json in sub json2perl.
This allows graceful death when input json is not valid.
0.13 2023-03-11
In the previous version I had forgotten an erroneous INSTALL_BASE
in Makefile.PL which pointed to $ENV{HOME}/usr.
This resulted in installing this module in installing user's
home dir, under usr/
Now this is fixed. Those who installed the previous version (0.12),
please make sure to uninstall it from $ENV{HOME}/usr.
0.14 2023-03-11
See 0.13.
0.15 2023-04-04
Changed the minimum Perl version required to be 5.008
in all test files and the Makefile.PL.
Aesthetic changes in Makefile.PL.
NOTE: this version has never been released to CPAN.
0.16 2023-05-30
1. Added a CAVEAT/WARNING in the documentation
for dump2perl($input) saying that it does
eval($input) which is a security risk if $input
comes from untrusted sources (e.g. user input).
2. Added a print statement to STDERR in sub dump2perl()
warning about said eval($input) (using a warn '').
This warning also states that dump2perl() will be
removed in the future because of this eval().
3. Added extra tests in t/10-scripts.t to check
that scripts fail when they should. In addition
to testing scripts that they succeed when they should.
0.17 2023-06-02
Requires Test::Script v1.28 because Test::Script::script_fails()
used by t/10-scripts.t was introduced there.
0.18 2023-09-01
Added two extra functions: jsonfile2perl() and yamlfile2perl()
which read JSON and YAML from a file into a perl data structure.
Added a new test file '12-from-file.t' which tests file-related
functionality including said two functions.
Documentation supplemented.
0.19 2023-10-03 01:00
added printing the yaml structure in test file '12-from-file.t'
if it fails to ease debugging.
0.20 2023-10-03 14:00
YAML::Load complains for strings like 'abc\"xyz',
so I have decided to remove all such strings from the
randomly created data structures of test file '12-from-file.t'.
This does not affect the behaviour of yaml2* subs but
only limits the testing range for input strings to yaml2* subs.
0.21 2023-10-03 14:45
Fixed bug in script/yaml2perl.pl to print the data structure.
0.22 2023-10-06 11:00
Switched from YAML to YAML::PP because the former failed on
some tricky input, see https://github.com/ingydotnet/yaml-pm/issues/224
Added extra test file (t/13-yaml-tricky-cases.t) for testing
these cases.
Added extra author test suit (xt/deficiencies run with make deficiencies)
to test the tricky cases for all YAML, YAML::PP, YAML::XS
Enclosed within eval {} all JSON::* calls (which were not already enclosed)
0.23 2023-10-11 11:30
Modified t/12-from-file.t to delete temp dirs on success
but keep them on failure.
I am trying to debug t/12-from-file.t killed on some test environments.
0.24 2023-10-12 12:00
String input to yaml2perl() is now being untainted in order
to avoid a bug with Perls < v5.14 which causes YAML::PP to
loop indefinetely and the OS to kill the script. Until YAML::PP is fixed.
0.25 2023-11-07 12:00
Change untainting yaml string input from ($yaml_string) = each %{{$yaml_string,0}};
to ($yaml_string) = keys %{{$yaml_string,0}}; because it was
warning about "each on anonymous hash will always start from the beginning"
Also a new test file was added (t/14-yaml-tainted-input.t) to
test yaml2perl with tainted yaml string as input.
0.26 2023-12-29 12:00
Clarified the documentation that internally YAML::PP is used and
that it has no problem with handling yaml strings which contain
escaped quotes.
0.27 2024-03-14 11:00
Chnaged the shortcut command-line interface
for the provided scripts so that
-r is equivalent to --terse and -d is equivalent to --indent.
Replaced 'warn()' with 'print STDERR'.
In eval()'ing code which dies (e.g. my $res = eval { YAML::PP::Dump($pv) }; ) I have used this 'if( $@ || ! defined($res) )' which now became if( ! defined($res) )'.
0.28 2024-04-08 18:00
Test file "t/12-from-file.t", at very few times, fails
(under M$) because comparing a data structure fails on
comparing two floats which must be the same and are the
same except the last decimal digit. As from this version,
the floats in the test data for this particular test is shorten
to max 4 decimal places.
These subs: dump2perl, dump2json, dump2yaml and dump2dump
are, from now, not exported by default neither they are
part of any export tag-group (they used to be in the ':dump' tag-group)
These subs must now be explicitly imported.
The reason is that 'dump2perl' (and the rest use this)
is eval'ing its input which may come from user.
Eval'ing (unchecked) user input is considered a security problem.
These subs will not be removed but be imported
by specific caller request.
Test "t/04-dump2-use-is-explicit.t" has been added to check that
these subs are not exported by default.
0.29 2024-10-15 18:00
Test file "t/12-from-file.t", still fails because of the issue
mentioned Changes of v0.28. Therefore I have reduced the number
of digits to zero. This is a minor issue and spent more energy than
necessary for it.