amorfo77/json - json - Gitea: Git with a cup of tea

2 Commits

Author SHA1 Message Date

Author	SHA1	Message	Date
Michael Gmelin	05b27e83b7	Exclude bytewise comparison in certain tests. These tests never worked - they weren't run before `d5aaeb4`. Note that these tests would fail because of this library ordering dictionary keys (which is legal). So changing the input files (or modifying stored cbor/msgpack/ubjson files) would make the tests work and they could get removed from "exclude_packaged". Also move parsing of files in these unit tests to within the inner sections, so that they're only parsed number_of_files * number_of_sections instead of number_of_files * number_of_files * number_of_sections (so, instead of close to 100k parses about 700).	2018-07-29 14:52:52 +02:00
Michael Gmelin	d5aaeb4cce	Make section names unique in loops, as catch doesn't support duplicate sections, see also https://github.com/catchorg/Catch2/issues/816#issuecomment-278268122 As a result, when built with gcc, loop iterations were skipped. When built with clang, the test aborted with an assertion in catch.hpp line 6222. This also addresses the issues discussed here: https://github.com/nlohmann/json/issues/1032#issuecomment-378707696 and here: https://github.com/catchorg/Catch2/issues/1241 Please note that this introduces new problems, as some of the unit tests fail now - the library stores keys in lexographical order, while the cbor/msgpack/ubjson examples store them in original order.	2018-07-29 10:44:36 +02:00

05b27e83b7

Exclude bytewise comparison in certain tests.

These tests never worked - they weren't run before
d5aaeb4.

Note that these tests would fail because of this library
ordering dictionary keys (which is legal). So changing the
input files (or modifying stored cbor/msgpack/ubjson files)
would make the tests work and they could get removed from
"exclude_packaged".

Also move parsing of files in these unit tests to within
the inner sections, so that they're only parsed
number_of_files * number_of_sections instead of
number_of_files * number_of_files * number_of_sections
(so, instead of close to 100k parses about 700).

2018-07-29 14:52:52 +02:00

Michael Gmelin

d5aaeb4cce

Make section names unique in loops, as catch doesn't support duplicate

sections, see also https://github.com/catchorg/Catch2/issues/816#issuecomment-278268122

As a result, when built with gcc, loop iterations were skipped. When
built with clang, the test aborted with an assertion in catch.hpp
line 6222.

This also addresses the issues discussed here:
https://github.com/nlohmann/json/issues/1032#issuecomment-378707696

and here:
https://github.com/catchorg/Catch2/issues/1241

Please note that this introduces new problems, as some of
the unit tests fail now - the library stores keys in
lexographical order, while the cbor/msgpack/ubjson examples
store them in original order.

2018-07-29 10:44:36 +02:00