Allow mocking a function that isn't linked into the exe

[ugoren]

When linking a test executable, we sometimes don't want to link it with all the libraries that are used in production. Mocking is very useful for this - if you mock all functions in a library, you don't need to link with the library itself. But C-mock searches for the real function with dlsym, and fails if it doesn't exist. So you have to link with it.

In my company linkage is a mess, so we have a local change that checks if the function exists only if we actually try to call it. This is very useful for us.
You can see the code in a branch in my clone.

But this changes behavior unconditionally, so I don't think it's appropriate for merge.

The only difficulty is how can we make it conditional. There are may alternatives, such as:

• Compile time switch (e.g. make ALLOW_MISSING=1).
• A separate set of macros (e.g. CMOCK_MOCK_FUNCTIONn_ALLOW_MISSING).
• When using a mock class, a variation of CMockMocker (e.g. CMockMockerAllowMissing).
• A macro defined by used before defining mocks (e.g. #define CMOCK_ALLOW_MISSING 1).

Can we add this functionality?

Thanks.

[hjagodzinski]

Sorry for late reply. I would opt for alternative approach. A lazy loading of a real function. As long as a given function is not mocked no error will be thrown even if it's not available.

[ugoren]

This is actually what we do in my company.
But can it be the default behavior, or should the user choose between this and eager loading?

[hjagodzinski]

A situation in which you call a function which you neither linked with or mocked I see as a bug. I would make a lazy loading the default behaviour.

[ugoren]

Thanks. I'll prepare the code and submit a pull request.

…

On Mon, May 7, 2018, 2:33 PM Hubert Jagodziński ***@***.***> wrote: A situation in which you call a function which you neither linked with or mocked I see as a bug. I would make a lazy loading the default behaviour. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#11 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AQ8EVh18S7twlA9iOhS1_wph7_YZVO_Jks5twDD-gaJpZM4TX74-> .

[hjagodzinski]

👍

[ugoren]

Submitted pull request. CI passed. BTW, C-Mock now fails to compile with latest Google Mock. The problem is with the old macros, which rely more heavily on Google Mock internals than the old ones.

…

On Mon, May 7, 2018 at 2:52 PM, Hubert Jagodziński ***@***.*** > wrote: 👍 — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#11 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AQ8EViwFAQahJumerDgQKpBOG1Yj3SHyks5twDVngaJpZM4TX74-> .

[hjagodzinski]

Can you create an issue on that so we can address it as a distinct problem?

[ugoren]

I'm not sure what you see as a distinct problem.

…

On Mon, May 14, 2018, 11:38 AM Hubert Jagodziński ***@***.***> wrote: Can you create an issue on that so we can address it as a distinct problem? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#11 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AQ8EVh_8YNyWgAR1hUojit_clLTKPlqYks5tyUKQgaJpZM4TX74-> .

[hjagodzinski]

I have in mind C-Mock now fails to compile with latest Google Mock.

[ugoren]

Ah, sure

…

On Mon, May 14, 2018, 12:09 PM Hubert Jagodziński ***@***.***> wrote: I have in mind *C-Mock now fails to compile with latest Google Mock.* — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#11 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AQ8EVn1Magr6j3fTC2A4rIYDVGbBOlfXks5tyUnkgaJpZM4TX74-> .