I think the generated file is too big about gpb HOT 7 CLOSED

0.0196?

There are a few options already to reduce the size of the generated code:

• {gen_verifiers,false}
• {gen_introspect,false}

This would leave only functions needed for encode and decode. Is this what you are looking for?

from gpb.

0.0196?

There are a few options already to reduce the size of the generated code:

• {gen_verifiers,false}
• {gen_introspect,false}

This would leave only functions needed for encode and decode. Is this what you are looking for?

{gpb_opts, [
{i, "priv/protobuf/"},
{module_name_suffix, "_p"},
% {o, "path/to/out_dir"}, %% both .erl and .hrl are generated here
{o_erl, "src/game_protobuf"},
{o_hrl, "include/"},
{strings_as_binaries, false},
{packed, true},
{gen_introspect,false},
{gen_verifiers,false}
]}.

Is this the configuration? In rebar3 config

But it doesn't seem to have taken effect, and the generated file hasn't changed

from gpb.

For me, your configuration works, and the options do make a difference. Did you clean first? Maybe it will not rebuild if only the options have changed.

from gpb.

For me, your configuration works, and the options do make a difference. Did you clean first? Maybe it will not rebuild if only the options have changed.

I confirm that the configuration has taken effect, and I have changed the file suffix to generate a different file.
Only one protocol generated over a thousand lines of code

from gpb.

In general, the layout of the generated code is the way it is in order to exploit as much as possible of the optimizations that the Erlang compiler and beam vm can do. It would likely be quite a bit of work to add an option to make the generated code smaller, plus there will be more maintenance. What is the reason why you want the code to be smaller? It is unlikely that I will spend an effort just based on opinions.

from gpb.

In general, the layout of the generated code is the way it is in order to exploit as much as possible of the optimizations that the Erlang compiler and beam vm can do. It would likely be quite a bit of work to add an option to make the generated code smaller, plus there will be more maintenance. What is the reason why you want the code to be smaller? It is unlikely that I will spend an effort just based on opinions.

Because the protocol files I generate when using other protocols are relatively small, I am not used to using GPB because my project generates a large number of protocol files. If the protocol files can be smaller, it can reduce the project's compilation time.

Thank you very much for your answer. :)

from gpb.

Thanks for the explanation. If the are many imported .proto files, then you can reduce the number of compiled files, and thus the compilation time. It is sufficient to generate code only the top-level .proto files, ie only for those that are not imported by some other. For example if a.proto imports b.proto and c.proto, then generate for only the a.proto. You can do this with the f option in rebar3_gpb_plugin: {f, ["file1.proto", "file2.proto"]} - allows you to specify a subset of all proto files found. But it will unfortunately be a bit large.

from gpb.