# Scoped enable\_language

**URL:** https://discourse.cmake.org/t/scoped-enable-language/766
**Category:** Code
**Created:** [March 4, 2020, 9:15pm UTC](https://discourse.cmake.org/t/scoped-enable-language/766 "2020-03-04T21:15:59Z")
**Posts on this page:** 3
**Page:** 1

<div class="post-metadata">

### Author: ![eriols](https://discourse.cmake.org/user_avatar/discourse.cmake.org/eriols/32/945_2.png) [@eriols](https://discourse.cmake.org/u/eriols)
#### Post date: [March 4, 2020, 9:15pm UTC](https://discourse.cmake.org/t/scoped-enable-language/766/1 "2020-03-04T21:15:59Z")

</div>

I have a project in language A with most toolchain files also in A.  
If built with unittests then there is one or two subdirectories in language B included.  
One single toolchain is also in language B.  
Depending on an option set there might also be some files in language C included which require a locally scoped enabling of language C…

How do I set enable\_language so that A is enabled everywhere whereas B is only enabled in the unittest subdirectory and/or for the particular toolchain (preferably without having an ugly if(unittests OR my\_special\_toolchain) clause)?  
I particularly want to avoid having to specify language B compilers for toolchains that do not use it which seems to be the case when all three are defined on the project command.  
I have tried a bit with only setting A on the project command and adding the others as need arise but I’d love some guidance on best practices/pitfalls to avoid.

Thanks.

---

<div class="post-metadata">

### Author: ![Angew](https://discourse.cmake.org/user_avatar/discourse.cmake.org/angew/32/229_2.png) [@Angew](https://discourse.cmake.org/u/Angew)
#### Post date: [March 5, 2020, 7:55am UTC](https://discourse.cmake.org/t/scoped-enable-language/766/2 "2020-03-05T07:55:54Z")

</div>

I have no experience with toolchain files, so I can’t comment on that part, but I can share a few caveats I’ve learned when working with `enable_language()`.

First off, the command may only be called _once_ for each language in the entire CMake run. So if you e.g. have two sibling directories which need language B enabled, you have to call `enable_language()` in a _parent_ of those two directories.

Second, the command must be called at file scope, not in a function, because it sets many variables for the language in its current scope.

However, there is no problem with having `enable_language()` controlled by an `if()`. Based on your description, I would probably structure the flow like this:

In the top-level CMakeList, do:

```auto
project(whatever LANGUAGES A)

if(the_one_toolchain_file)
  enable_language(B)
  set(enabled_B TRUE)
endif()

```

In parent directory of all unit tests:

```auto
if(not enabled_B)
  enable_language(B)
  set(enabled_B TRUE)
endif()

```

In a directory above all C language sources:

```auto
if(the_option)
  enable_language(C)
endif()

```

---

<div class="post-metadata">

### Author: ![eriols](https://discourse.cmake.org/user_avatar/discourse.cmake.org/eriols/32/945_2.png) [@eriols](https://discourse.cmake.org/u/eriols)
#### Post date: [March 5, 2020, 8:05am UTC](https://discourse.cmake.org/t/scoped-enable-language/766/3 "2020-03-05T08:05:42Z")

</div>

Thanks, this is helpful! So to get some sort of scope you need to do a bit more file separation.

An alternative that could be of interest is if there is a convenient way to turn off the requirement to include a compiler for a certain langague that’s enabled above.  
Eg cmake complains loudly when crosscompiling if I don’t do a set compiler for the enabled language B that I don’t (normally) use in that case.