parts/inc/threads: Add hints

(cherry picked from commit 5bc28160c8a5ea859054713667f051e47e46802d)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/mess: Don't 'provide' a porting test macro

This macro is not for public consumptions, but only for porting tests.
Change so this file doesn't cause D:P to provide it.

(cherry picked from commit fdb67922468b8439f47c1c80bba2ad557bec5dc3)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/mess: Adjust prototype to match blead

(cherry picked from commit 129ae046bd633f28af9e86ab704b11d7342e2137)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/magic: Fix to run on early perls

No for postfix modifier; use numeric comparison for $]

(cherry picked from commit c0b235db11fc135c6da5f5b2890fa44bc1aad10c)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/magic: Adjust prototype to match blead

(cherry picked from commit 692ec66c0bf83f43232d8a5b52d582346a6948d7)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/variables: Grammar in comment

(cherry picked from commit b38f592c79e9cb96807bf04e902d04f5004a7914)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/uv: Adjust prototype to match blead

(cherry picked from commit b0b2502b166d7e41040ad14b557335014ded7f7c)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/uv: Use > to compare $]

(cherry picked from commit a54aebf6bde1ad7510f02a2e83b35658973c8994)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/utf8: Symbols are not really for external use

This used __UNDEFINED__ to define a bunch of constants.  However, there
is nothing backported that allows the user to specify any of these, so
there is no need for them to be publicly accessible.  And, if something
is important enough to be backported, it ought to be documented.  These
symbols should not be documented, so __UNDEFINED__ is not the proper way
to deal with them.

(cherry picked from commit 359209af4e95548cbd7291128549ce2071e8e555)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/utf8: Add hint for utf8_to_uvchr()

And a Replace pragma

(cherry picked from commit 56f6e61be8269a8a7959eae3e75dd8fbd2976aa6)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/utf8: Fix typo in comment

(cherry picked from commit d93ef4febe6f8e62de50dfe22f0b9c80b47ec967)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/utf8: Fix hints for is_ascii_string

This hint was the wrong thing.  We really want it to come out for the
API functions we want to caution about.

(cherry picked from commit 4931fb4d1f76616701bcb786949827409cde1e09)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/utf8: Use numeric comparisons with $]

The string ones failed on 5.4.3

(cherry picked from commit 961d99e3be46e3bca016afb899440028e7e734e6)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/utf8: Fix for loops to work on early perls

You can't use a for statement modifier, nor declare a variable as part
of the for statement.

(cherry picked from commit 5ebfad4758132f28a11b3a5c1e58c9c0c8bc3a30)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/utf8: Fix utf8_to_uvchr_buf for 5.7.0

It turns out that 5.6.1 is more advanced than 5.7.0.  So this commit
uses a further check to decide if it can use a substitute function that
provides decent error checking, or if it has to fall back to one that
doesn't.

(cherry picked from commit 8e3920d656441ef0ba0a5dcef98c1064aab26412)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/utf8: make sure var is def'd before using

I'm not sure that this makes a difference, but it's cleaner to not

 #define foo bar

if bar isn't defined.

(cherry picked from commit 686eafa088d7b41bc630e6889cbe546a46e8b366)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/utf8: Wrap qr in string eval

Very early perls fail to compile this because qr doesn't exist and is
viewed as a syntax error, even though the tests don't get executed.

(cherry picked from commit 682fa53f57973b5a263c3b6742a54fa90d7995b6)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/sv_xpvf: Adjust prototype to match blead

(cherry picked from commit b022731d60d59e7719049671ae04d798158205c9)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/Sv_set: Fix to work with early perls

You can't have a for () modifier after a statement, and need to use
numeric comparisions with $]

(cherry picked from commit 4ab672ea5af5ec2a1e561d02941b4582d68dad6c)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/sprintf: Add Warning, Replace: pragma

This really shouldn't be used; instead snprintf()

(cherry picked from commit c36539f5e553c48412c1d4a770359097c3070f6e)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/shared_pv: Adjust prototype to match blead

(cherry picked from commit 3346d69075fdaaa1344ececa15483c0ccb9ec170)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/misc: Remove extraneous ord() calls

(cherry picked from commit 0257260a82e7924470d25bd227890d991a973fb5)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/misc: Shorten sub names

These all had an unnecessary prefix that was to prevent name space
collisions, but the infrastructure of D:P already took care of that.

(cherry picked from commit d8d82913b626ad2f6c5179d4a95b26fe52048a70)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/misc: Can't use GCC brace groups before 5.4

(cherry picked from commit b036d9249ac2b7f64d8a3e9b2f520a469f86fcdb)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/misc: Change Hints into comments

I wrote these not knowing what Hints were for (since clarified in
HACKERS).  They are really comments for a person reading this file.

(cherry picked from commit 0dcefa7708a4113757a215c3e5386b2ead255135)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/mess: Use numeric comparisons for $]

These give better results, and the string compares were broken on 5.4.3.

(cherry picked from commit 01fbc5c65146c6cce5845f6accca6515ca4509b3)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/mess: Don't use 'for' modifier

It doesn't work on early perls.  Instead wrap with a for loop

(cherry picked from commit c87fc33e0370528f8e0376e48ef0ea4f82395afb)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/mess: Some things are broken in the 5.7 series

while working earlier and later

(cherry picked from commit fa7ee13073f7d45e205cd6cdd7e727c83739ed30)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/mess: Some things don't work before 5.4.0

(cherry picked from commit d2544abef917c023c211f4e084ff464c0e07b822)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/mess: \N{} broken in more versions

Skip these tests on more versions than before

(cherry picked from commit f62175a208c31a08d85fb2d73b7a6d496b04ad37)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/cop: Adjust prototype to match blead

(cherry picked from commit c19b8832d23fd6999231e451e86a50fb43720790)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/call: Fix to run on old perls

Use numeric, not string, comparisons with $], and can't have postfix
'for' modifiers

(cherry picked from commit 19e824df0bd6883d8f3deef860bbaf40249d811b)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/call: Adjust prototype to match blead

(cherry picked from commit 444c878fc618827c1d33c2257476c31815c50eda)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/format: Fix to work on early perls

It just isn't ported before 5.4, and postfix 'for' isn't supported

(cherry picked from commit 88a327df23bd4647730a8a6bc2e99879a1f96fc9)
Signed-off-by: Nicolas R <atoomic@cpan.org>

ppphtest: Update based on previous commits

The wording of some messages has changed.  This also adds some test
names

(cherry picked from commit 80d37f0acfdb10cf180655b4accf49ea09e93379)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/mkapidoc.pl: Rewrite entirely in perl

The old version, partially in sh, grabbed every file in the directory
structure, looking for apidoc lines.  Unfortunately that included
things, in .git, for example, and those could be out-of-date entries,
with the result that there could be multiple entries with different
prototypes for the same macro.

This commit changes to read only MANIFEST files, and a few generated
files.

(cherry picked from commit d8bc0a3f40e83e52934696240c2d89398465a56c)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/ppphdoc: Add caution to pod

(cherry picked from commit 35558be45ff5f20805c6ea56f66b900231ac4e33)
Signed-off-by: Nicolas R <atoomic@cpan.org>

PPPort_pm.PL: Silence warnings on early perls

Enclose some hash keys in quotes.

(cherry picked from commit cef8c4f4fa7a02a330ed79ac67934ad32e5e48e0)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/regenerate: make sure the .fnc files don't overlap

If they do, this could show that they weren't recently regenerated, or
that there is a bug in the regeneration code.

(cherry picked from commit 51e09cbdf5df04a4e2a7304c5a5bf39f9e91ff00)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/scanprov: Store information about all fcns

This makes it convenient for someone to find out in what version
something got added.  I'm not publicizing this just yet, and
ppport.h --api-check won't return it for non-public ones at the moment,
but the information is now there for those who know where to look, and
perhaps for future public consumption.

(cherry picked from commit a9fc5abd7e81911cb92ed3554300c340a8f3c88e)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/apicheck.pl: Use the passed NEED_foo macros

These no longer need to be hardcoded in two places when adding new ones.

(cherry picked from commit 867650c3a523c9e8f5014ba02ef52e88692a5302)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/regenerate: get a list of all NEED_foo macros

And put it in the environment for apicheck.pl.  This will keep a
maintainer from having to remember to add these in two places.

(cherry picked from commit 6734eaffe2a3aedd68a6c610a0f0258c2df850e9)
Signed-off-by: Nicolas R <atoomic@cpan.org>

RealPPPort_xs.PL: Fix spelling in comment

(cherry picked from commit acd6caec2173ad3c12f04c788b8b4ba64746e2b1)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/regenerate: Add some error checking

Make sure that things we skip testing are on the list of such things,
which has information that we otherwise would glean from testing

(cherry picked from commit 98ff9eaf3c08405d27ba8380835087116e7c98f4)
Signed-off-by: Nicolas R <atoomic@cpan.org>

Update parts/ppport.fnc

(cherry picked from commit 09bc9a820d17f1010e8b6bacdf1e6d319ef4e41b)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/apicheck.pl: make macro public in X+M flagged functions

The X says we export a function even though it's not public API, but the
M says there is an extra, public, macro that maybe used to call the
function, but now calls something else, and that macro is public, so
should be tested.

(cherry picked from commit 7528b129460c777edc99db3e5555b38a41a9390f)
Signed-off-by: Nicolas R <atoomic@cpan.org>

Create regeneration files in final directory

Prior to this commit, the files for what happened when were placed in
the parts/todo directory, then moved to parts/base.  Now they go
directly into their final resting place

(cherry picked from commit 18edcc361cd5a2a208ae85a627c3c861c5ce3d74)
Signed-off-by: Nicolas R <atoomic@cpan.org>

Pass values from mktodo.pl to apicheck.pl

This uses an environment variable to send the version number across a
make to apicheck.pl, where it is inserted as comments in apicheck.c.

I found that when debugging, it is useful to have this information
handy throughout the file.

(cherry picked from commit 2a055a1a8526c1b721d10078ce9bf39178c7ea62)
Signed-off-by: Nicolas R <atoomic@cpan.org>

ppport.fnc

(cherry picked from commit 139ab76c6ce97398246dda402e0e98ac32e1df1d)
Signed-off-by: Nicolas R <atoomic@cpan.org>

PPPort_pm.PL: Add __UNDEF_NOT_PROVIDED__

This is like __UNDEFINED__, but does make "provided" the macro it
targets

(cherry picked from commit 63a8d75f3253d340ded4bbbc4246d68fdf7ed0e7)
Signed-off-by: Nicolas R <atoomic@cpan.org>

PPPort_pm.PL: Add an item for unknown min supported version

The pod for this has sections of what elements are first supported when.
Unfortunately there are some that we do not have that information.

The cure for this is for someone to contribute pod for that element to
the perl core, but in the meantime, this lists those we couldn't figure
out.

It uses ppport.fnc to figure out many of the ones it otherwise couldn't.
And it is possible to enhance this to figure out more.  But just
documenting them removes them from the list.  And they should be
documented anyway.

(cherry picked from commit 5b548c97ec59bfb65f3caefea9fcf50cf8cd6d49)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/ppptools.pl: Make sure mathoms fcns are deprecated

This makes sure that we consider functions in mathoms as being
deprecated.  The current blead regen/embed.pl already does this, but
this makes sure, should that get reverted.  A mathoms function with
an M flag indicates there is a macro by that name that avoids the
function (which must begin with 'Perl_'), and the macro isn't deprecated
unless the D flag is also present.

(cherry picked from commit 8ed3f9e76aca7a9341a8878d2d7c61a8b8717621)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/apicheck.pl: Don't test macros with weird params

These are marked for us in the =for apidoc line.

(cherry picked from commit beb801439ddfa3dc555a1ea0ff5a1794715b6ca4)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/ppphbin: Use dependency info for min support

Some items don't get tested, and so we don't know their minimum
supported version.  But if they're dependent on some other thing whose
minimum version we do know, we can glean that they aren't supported
prior the the dependent thing's support.

(cherry picked from commit 311664e55550bf20d255a952ad2532c1c8747c73)
Signed-off-by: Nicolas R <atoomic@cpan.org>

Add script to regenerate ppport.fnc

Now, instead of this being static, it will be regenerated.  This allows
us to create tests for macros that otherwise wouldn't be.

(cherry picked from commit b730918253cf5ea4167f6c3ee7096dd2e4068506)
Signed-off-by: Nicolas R <atoomic@cpan.org>

PPPort_pm.PL: Handle blead specially

blead actually shouldn't be reported, as the symbols it shows as
existing are (due to the algorithm) actually ones that don't exist in
any version.  By ignoring it, we omit wrong information.

(cherry picked from commit 97d35610a4fccf694db9a601b102b5a2888b9713)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/uv: Add Replace: pragma

(cherry picked from commit fef460fa0e876800c7cde33b188dd2c95a6ec231)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/SvPV: Add Replace: comments

(cherry picked from commit 0b4cfb7fef017e2bb9ff99e223a6fce5f2bb7794)
Signed-off-by: Nicolas R <atoomic@cpan.org>

ppport.h: Make use of scanprov's new returns

(cherry picked from commit 0e1a8da5dde291aa56a336ed14ce721f3e6c028f)
Signed-off-by: Nicolas R <atoomic@cpan.org>

ppport.h: Reword --api-info entries

This makes the wording, I believe, more succinct and friendlier.

(cherry picked from commit 272e25aaf16afedb1ad27079644861e472aa58bc)
Signed-off-by: Nicolas R <atoomic@cpan.org>

ppport.h: --api-info Reorder output of hints/depends info

It looks better after everything else

(cherry picked from commit 0b98f2fcb18283bc726c16fc45ba8e8efcf8017c)
Signed-off-by: Nicolas R <atoomic@cpan.org>

ppport.h: Suppress useless warning

If an element is ported as far back as we know about, there is no point
warning about that.

(cherry picked from commit cb6834059fb808db043092b42e421a66023ea26a)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/ppphbin: Do calc once, and store in variable

(cherry picked from commit ef5acaa4d66699bbf87a9051192c1db82f79d406)
Signed-off-by: Nicolas R <atoomic@cpan.org>

PPPort_pm.PL: Make use of scanprov's new flags

scanprov now gives us a type for the symbols it adds.  Pass that type
along to ppport.h

(cherry picked from commit ba8e4f38988bca4a29edeb751b6c461eb671864b)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/scanprov: Scan for functions we didn't test

mktodo.pl tried to compile each element it knows about.  But not all
will actually compile, as some compile only when certain Configure
options are specified on certain platforms, like ones dealing with quad
math, or GLOBAL_PRIVATE_STRUCT.  But those functions are in embed.fnc
and the hdrs that get generated from that.  We can use the same
mechanism as elsewhere in scanprov to look for the first mentions of
these.

(cherry picked from commit 8530a7806b5283c981eb3e22ba311be35924a071)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/scanprov: Extract code into a function

This is in preparation for it being called a 2nd time.

(cherry picked from commit be556cbfa5105ba27a1099ca21625c4958ee8992)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/mktodo.pl: Get working on threaded perls.

If we don't ignore issues with this libc symbol, pthread_getspecific(),
we don't work on threaded perls.

I don't understand why this is undefined.

(cherry picked from commit dd767ef602ff590b061b8d54d83eaec1945ec473)
Signed-off-by: Nicolas R <atoomic@cpan.org>

ppport.h: Turn off some likely warnings

I'm not sure when this ability to turn off warnings was introduced, but
putting it inside a block eval should avoid any issues.

(cherry picked from commit d1378eaded6746a4c36230f9eb803c2bce7edb3d)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/apicheck.pl: Don't skip some cases

We only should create a single test for a given function.  But if that
function is like this:

  #if foo
    int bar(...)
  #else
    char bar(...)
  #endif

We would only generate a test for the 'foo' case.  Uniqueness of a
function includes the compile time conditional expression for it.

(cherry picked from commit 2d6698cc1724195e18acd67fb8fddf9527372eda)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/scanprov: Add in exceptions

Things like RETVAL were being listed as not having any portability
information, whereas they are actually valid for all current releases.
Elements like these can't be tested individually because they are too
tightly bound with XS, which is used to do the testing.  Therefore, they
are in a list of exceptions.  This commit causes scanprov to use this
list and insert the items into the proper version files.

(cherry picked from commit 06245c0f864cb3b530028a3fbb2baa3e540cccbf)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/apicheck.pl: Extract list into lib fcn call

This is in preparation for scanprov to use the same list.

(cherry picked from commit b3b8255b1cb743f160e6d9f33fa52b7b7149d53b)
Signed-off-by: Nicolas R <atoomic@cpan.org>

PPPort_pm.PL: Revise pod

(cherry picked from commit 7c5cbcc2317710ec510aad42e8c64ca65ca0f7a4)
Signed-off-by: Nicolas R <atoomic@cpan.org>

XXX parts/inc/ppphtest: Update

Changes in this distribution have caused certain things to change
slightly.  Update so the tests pass again.

(cherry picked from commit 53824f57c7b96239c40fa20a89f6b184cca12b77)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/scanprov: Add a code letter to generated lines

This is in preparation for other types of lines being handled by
scanprov.

The lines in parts/base that come from mktodo.pl have the codes that nm
outputs for each symbol, plus some internal codes.  This adds M, which
doesn't appear to be an nm code.

(cherry picked from commit cb5fef4b862d528eba649ee18870324ac26a4e68)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/scanprov: Extract code into a fcn

This is in preparation for it being called from other places

(cherry picked from commit 56bf67b75f766a453aa2373db9f6f09935cbae2e)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/scanprov: Ignore comments in scan

As noted in the comments in scanprov, the results could be thrown off by
something being mentioned in a comment in a scanned hdr.  This commit
removes the comments before scanning.

I'm unsure of what else could throw off the results, but I can imagine
there is something that I (and previous maintainers) haven't thought of.

(cherry picked from commit 413b3996cf5d19cc43357ba714b0662c93d73903)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/scanprov: Restrict symbols to identifier-like

Previously, it used \w+, which gathers things like numbers.  We are
instead interested in identifiers.  It probably doesn't make a
difference except that we don't needlessly add a bunch of things to the
generated hash

(cherry picked from commit 83a7475bfcfa1bd2d56b4cb1a77f43e7059c2f20)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/scanprov: Handle blead properly

Previously, the code just removed blead from the list of perls it dealt
with.  But this doesn't work, because the algorithm (heuristic,
actually) relies on going through all the versions, and the symbols
erroneously ascribed to blead will instead erroneously be ascribed to
the next lower version.  Instead, this commit changes so that blead is
simply not written.  (The comments detail why the symbols are
erroneously ascribed to the highest version checked.)

(cherry picked from commit 89c7fcf8b81b3390a318abd2d2f5875305fc5fd1)
Signed-off-by: Nicolas R <atoomic@cpan.org>

PPPort_pm.PL: Rename variable for clarity

Doing so showed that we already had an array with what we wanted, no
need to go through a hash again.

(cherry picked from commit 4f91b5dc1d8171a6bc6b54fb8d76fa6bca697f7b)
Signed-off-by: Nicolas R <atoomic@cpan.org>

PPPort_pm.PL: Comments, white space

(cherry picked from commit c9bb660cea95c2c98a0f5a6a63b7f5fdf1d2fadc)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/ppphbin: Change variable name

This makes it clearer at first glance as to what is going on

(cherry picked from commit 38edf48e7b91ba30e1c26958445deac475d52347)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/ppphbin: Add comments

(cherry picked from commit 9e8dd684615d93294c81a4bff7e7610b0621d1fe)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/ppphbin: Fix accumulation of Hints/Warnings

Prior to this commit, if any line in the middle of one of these didn't
begin with an asterisk (a typical comment continuation marker), the hint
was silently discarded.

(cherry picked from commit d7b17de7bd53bc7f5c58e0aebeec7420e0de860d)
Signed-off-by: Nicolas R <atoomic@cpan.org>

Rename some data files

These are renamed to their proper 6 digit perl version names.
Previously the final part was multiplied by 10.

(cherry picked from commit d48910fdea411975c3e0be496800f8ea73c69091)
Signed-off-by: Nicolas R <atoomic@cpan.org>

pppport: Print hints/warnings always.

Previously they were displayed only if somethng else was.

(cherry picked from commit 27527f7a456d30199751a43bdba64212daad7e61)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/ppphbin: Add clarifying text to output

(cherry picked from commit 1f910f5b55fe9bfb8d755dae343f2b785e44aee7)
Signed-off-by: Nicolas R <atoomic@cpan.org>

ppport.h: Don't say unsupported if avail in earliest version

If this API element is available in the earliest version we have info
for, then it actually is supported.

(cherry picked from commit 083f8295dac44481178034faba58f096dda25703)
Signed-off-by: Nicolas R <atoomic@cpan.org>

ppport.h: Don't print unsupported info for invalid entries

By 'invalid', I mean ones that are, say deprecated or inaccessible
outside core, etc.

(cherry picked from commit a114ba3327ec8379d16b870e91d260b1e99b1866)
Signed-off-by: Nicolas R <atoomic@cpan.org>

ppport.h: Print out more API info

This uses the flags from embed.fnc to enable printing better API info
for elements that are, say, deprecated or core-only.

(cherry picked from commit 1fc2b5279cfb31e354f3bba615df71f721ad071b)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/ppphbin: Change variable name

This was a smash up of several words that I kept reading wrong

(cherry picked from commit 20bf156f900f5a012fc3610ddc54db6092ed907e)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/ppphbin: Change option default

The verion that we default to shouldn't be earlier than the earliest
version we know about.

(cherry picked from commit ef70102090b532971146c8ea9d67df4369abb175)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/mktodo.pl: Don't discard earliest info

When regenerating the files of what came when, the code uses version n
as a base to see what version n+1 added.  Prior to this commit the
information from version 0 was simply discarded.  That meant you had to
have one more perl to compile than what you wanted info for.  That is
wasteful and unnecessary.

This commit changes so that a file is generated for the earliest perl
being generated, containing all the API symbols known to that perl.
Effectively, it assumes that all of them popped into existence when that
version came out.  Words will be changed in the output to indicate that
these symbols have been available AT LEAST since this version.

For many years, the earliest perl that we had info for was 5.003_07.
That meant that the regeneration was run with an even earlier perl.
With this commit, wherever we stop, we have information from it, and
don't have to throw anything away.

(cherry picked from commit 28c1bee885eb2373eda222e45b5af49cec6ccfb5)
Signed-off-by: Nicolas R <atoomic@cpan.org>

PPPort_pm.PL: Don't hard-code min/max versions

We can calculate this from the files we have generated, so manual
intervention is no longer required.

(cherry picked from commit f778898fbc24791c7b1181ac5e4d9b677d7e3233)
Signed-off-by: Nicolas R <atoomic@cpan.org>

Add an argument to mktodo.pl

This will be the directory in which to place the output files.

This will enable not having to write over one set of file as we go
along.  Previously everything got written to one directory, then the
files were moved, and the directory filled again.

This will also enable passing in info without having to add even more,
mostly redundant, arguments.

(cherry picked from commit 23e4148517314fb987306cc65255c25028174c16)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/mktodo.pl: White-space only

(cherry picked from commit e01f423f20117c9315f3b05bd485a78cab5b8212)
Signed-off-by: Nicolas R <atoomic@cpan.org>

parts/inc/inctools: Add function to format version

This takes a version number and creates a floating point version with
exactly 7 digits with a decimal point after the 5.  This is used as the
first line in files in parts/todo,base, as well as in the data in
ppport.h.

(cherry picked from commit fad2103a9a2c28ff439b9b7c851a9e09194e91a6)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/mktodo.pl: Use debug levels.

(cherry picked from commit cb2f9c687c7687295c5f63aca668edba9a395751)
Signed-off-by: Nicolas R <atoomic@cpan.org>

Change --debug option to have levels of detail

(cherry picked from commit d62bda55bbfa2de367141dce1774eed7db9c3b64)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/regenerate: Remove exists from if tests

I'm unsure as to why his is needed, but I found there were problems when
the code uses exists instead of just checking if the value is true.
This is a tiny hash, so autovivification isn't a performance issue at
all.

(cherry picked from commit 41457fa69251e60ddfce2e4f870a09a0115bfc6b)
Signed-off-by: Nicolas R <atoomic@cpan.org>

devel/regenerate: Add a few comments

(cherry picked from commit 13c691ffddb3eef81dc61589f8c1bba382c5a1f2)
Signed-off-by: Nicolas R <atoomic@cpan.org>