Added support for foo.some_thing === foo.someThing

[hason]

Alternative implementation of #934. Works well for methods, properties and array calls of objects with minimal performance impact. Results of twig benchmark (fabpot/twig-perf):

                               |       v1.12.3 |       v1.16.3 | origin/master | origin/methods | 
------------------------------------------------------------------------------------------------
empty                          |     13.5  0.0 |      8.8  0.0 |     14.4  0.0 |     20.5  0.0 | 
empty/B                        |      7.9  0.0 |     14.1  0.0 |     14.6  0.0 |     18.9  0.0 | 
simple_attribute               |     88.2  0.9 |     99.8  0.9 |     99.2  0.8 |     99.3  0.9 | 
simple_array_access            |    180.2  1.3 |    132.8  0.9 |    120.6  0.9 |    127.1  1.0 | 
simple_method_access           |     94.4  2.4 |     92.9  2.7 |    101.7  2.7 |    119.7  2.9 | 
simple_attribute_big_context/B |     90.5  0.9 |    101.3  0.9 |    100.5  1.0 |    116.1  1.3 | 
simple_variable                |    710.0  0.7 |    698.6  0.7 |    711.9  0.8 |    680.2  0.7 | 
simple_variable_big_context/B  |    702.2  1.3 |    668.3  1.3 |    795.9  1.2 |    674.3  1.2 | 
simple_foreach                 |     10.9  0.0 |     21.0  0.0 |     15.2  0.0 |     10.7  0.0 | 
simple_foreach/B               |     10.1  5.7 |     16.3  7.7 |      9.9  5.0 |     14.5  4.7 | 
empty_extends                  |      9.5  0.0 |     17.4  0.0 |     17.6  0.0 |     13.9  0.0 | 
empty_extends/B                |     16.1  0.0 |     17.5  0.0 |     10.2  0.0 |     22.6  0.0 | 
empty_include                  |     10.3  0.0 |     10.0  0.0 |     12.1  0.0 |      9.5  0.0 | 
empty_include/B                |     14.8  0.0 |     14.5  0.0 |     15.4  0.0 |     14.4  0.0 | 
standard                       |     22.1  0.0 |     12.6  0.0 |     12.9  0.0 |     14.0  0.0 | 
escaping                       |    101.8  1.2 |    110.9  1.2 |    107.5  0.8 |     99.2  1.0 |

[simensen]

Wow! This seems like it might be great. I don't know enough of the C code to know how well this is implemented. That is part of the reason I haven't had a chance to follow up on #934 yet. Thanks for running with this!

[stof]

wouldn't this change cause some issues with protected ? Note that I'm not sure about my comment as I'm not a C developer

[hason]

@fabpot ping

This PR does not hurt performance #934 (comment) and works with properties and arrays #934 (comment)

[fabpot]

I would remove this one.

[fabpot]

@tucksaun @jpauli Can you have a look at the C part of this PR?

[fabpot]

for the non-debug mode, what about keeping this cache on disk? Would that makes things faster? It looks like there is a lot going on in the getCacheForClass() method and I cannot believe that it does not have any impacts on real-world templates.

[hason]

Support for the cache may be added in another PR.

[lsmith77]

can we give this PR some life again?

[elnur]

😞

[fabpot]

This PR was auto-closed because I've removed the master branch and this PR was based on master. I cannot reopen it unfortunately.

[fabpot]

Changing the base branch to 2.x worked :)

[fabpot]

This one never moved forward as we did not get feedback on the C implementation. I propose to rebase it on 2.x where we don't have the C extension.

[hason]

@fabpot The C extension in 1.x is no longer used. Should I update this PR on 1.x or 2.x?

[xabbuh]

@hason The C extension is gone.