No problem, thanks for reporting, even if it turned out to be a non-issue :-)

#include <tao/pegtl/contrib/integer.hpp>
int main () {}

Gives

In file included from a.cpp:1:
../../../external/taocpp/pegtl/include/tao/pegtl/contrib/integer.hpp:63:60: error: ‘numeric_limits’ is not a member of ‘std’
   63 |       template< typename Integer, Integer Maximum = ( std::numeric_limits< Integer >::max )() >

Sorry, my bad I was really too far behind !

@Bjoe Thanks a lot, greatly appreciated!

@d-frey I created a proposal. See PR #233

This is a proposal to fix failing android CI builds. See issue #216

It uses a docker where the android NDK r22 beta 1 is installed to build PEGTL.

See also travis build: https://travis-ci.org/github/Bjoe/PEGTL/jobs/745332637

It needs maybe some "cleanup" regarding to the output.

Maybe some output of the current NDK/SDK version would be useful.

Sorry, I meant <limits> of course.

#include <tao/pegtl/contrib/integer.hpp>
int main () {}

Gives

In file included from a.cpp:1:
../../../external/taocpp/pegtl/include/tao/pegtl/contrib/integer.hpp:63:60: error: ‘numeric_limits’ is not a member of ‘std’
   63 |       template< typename Integer, Integer Maximum = ( std::numeric_limits< Integer >::max )() >

contrib/json no longer consumes the whitespace after the json value.

Your changes fixed my issue. Thank you.

@d-frey Thank you for condolence ... it's three months ago so we are getting back to normal...

I found some more information about the NDK r22 release. So there exists a NDK "22.0.6917172-beta1" version. There are also a changlog for this version.

I used this version locally to build PEGTL for android and it compiles ! :-)

I try now to add this beta version in travis .. see https://travis-ci.org/github/Bjoe/PEGTL/builds/744958133

It will help you to test if PEGTL is compiling for android. It won't help your "customer" as NDK r22 is not yet released.

To introduce -llibstdc++fs in the NDK, I do not recommend this. If you really will have "now" a solution for your "customer", then I purpose to use boost::filesystem. This works 100% on android. I already use boost::filesystem on android. But this means, your project is depended on a third party lib, not sure if you like it.

If recommended following: Use the NDK r22 "Beta" in travis, so you know that PEGTL will compile on android (I'm already on it). Add a hint for the android users that your project can compile/use on android when NDK r22 is released (or if they like use the beta NDK r22). I think it will not take a long time if NDK r22 is released.

@ColinH @d-frey As @wravery mentioned. std::filesystem is supported at NDK r22. At the moment NDK r21 is released. So r22 is a future release.

See also NDK issue: android/ndk/issues/1272

NDK r21 is not yet fully support C++17 features.

@d-frey Sorry for the late response... some terrible thinks happens on Aug. ... anyways

I'm surprise why the the android toolchain is using libstdc++ ... it is not anymore updated by the NDK. PEGTL should use libc++ instead. Give me some day, I will create a PR to fix it.

I was also really surprised @ColinH This file system is a (stupid) proprietary NFS that is used on my office network. It is very bad and probably very expensive... I have no control upon it, but obviously since work stations are Linux boxes, we have recent gcc or clang compilers running on them.

Are there any systems that have a C++17 compiler but don't support Unicode filenames?

I'm learning PEGs by trying out a basic pascal-ish language. I've got the following grammar:

struct Word: pad<sor<plus<identifier>>, space> { };
struct BasicReal: rep_min_max<1, 1, star<digit>, string<'.'>,plus<digit>> { };
struct Exponent: sor<string<'e'>, string<'E'>> { };
struct OptNegPos: opt<sor<string<'+'>, string<'-'>>> {};
struct RealDigits: rep_min_max<3, 3, digit> { };
struct ScientificForm: rep_min_max<1, 1, sor<rep_min_max<1, 1, Exponent, OptNegPos, RealDigits>>> { };
struct Real: pad<sor<BasicReal, rep_min_max<1, 1, BasicReal, ScientificForm>>, space> { };
struct Special: pad<sor<string<','>, string<'('>, string<')'>, string<'['>, string<']'>, string<';'>, string<'<'>, string<'>'>, string<'*'>, string<'/'>, string<'+'>, string<'-'>, string<'='>>, space> { };
struct Integer: pad<plus<digit>, space>{ };
struct HexInteger: pad<rep_min_max<1, 1, string<'0', 'x'>, plus<xdigit>>, space> { };
struct Grammar: must<opt<sor<bof, bol>>, until<plus<sor<Word, Real, Special, Integer, HexInteger>>>> { };

All my code does right now is tokenize the input, but I keep getting a parse error when I try to evaluate the following code:

program abc
integer a,b,c
real x
begin
    b=6;
    a=b+-4;
    x=-5.63e002;
end
end

It always breaks on the final line, either the "end" or, if I add an extra newline, on that newline. Is there something I'm missing?

For now we've kept the combinator table, albeit on the "getting started" page, but have removed the other tables since they would need to be greatly extended in order to be really useful, which then creates its own set of issues.

Yes, I have. I don't have any further questions at this time. I probably will sometime later though. :)

On 11/19/20, Dr. Colin Hirsch [email protected] wrote:

How are things going, have you managed to make some progress? Any further questions or issues that you might want to discuss?

-- You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub: https://github.com/taocpp/PEGTL/issues/227#issuecomment-730576767

-- Signed, Ethin D. Probst

@drtconway The grammar analysis now also supports the ICU rules.

How are things going, have you managed to make some progress? Any further questions or issues that you might want to discuss?

@kelvinhammond It sounds like what we internally call "switching style" might be useful for you, i.e. changing the States and/or Actions for different parts of the grammar, see https://github.com/taocpp/PEGTL/blob/master/doc/Actions-and-States.md#changing-actions and below.

@drtconway I'll look into adding grammar analysis integration for the ICU rules.

Do you have experience with the NDK? We would greatly appreciate a PR that would fix things again.

Coverage remained the same at 92.432% when pulling cc9742b992df740f16d46d2a2238740afc966b77 on catalogm:json_early into a6716edb39b822dc2ecf7d7b4769cf1611db7dcc on taocpp:master.

Tests pass, master fails because of build errors in the json examples. I'm not sure how to fix it. @d-frey can you take a look at the example::control in the json_build.cpp, etc examples?

I needed this so that I could parse an expression that takes json value or an identifier followed by one or more whitespaces and then a word but the json value was consuming all of the whitespaces.

contrib/json no longer consumes the whitespace after the json value.