# No SOURCES given to target: \_\_CMAKE\_\_CXX26 on ubuntu 25.04

**URL:** https://discourse.cmake.org/t/no-sources-given-to-target-cmake-cxx26-on-ubuntu-25-04/14240
**Category:** Development
**Tags:** os:linux, gen:ninja
**Created:** [June 16, 2025, 11:26am UTC](https://discourse.cmake.org/t/no-sources-given-to-target-cmake-cxx26-on-ubuntu-25-04/14240 "2025-06-16T11:26:10Z")
**Posts on this page:** 1
**Showing post:** 17

<div class="post-metadata">

### Author: ![vito.gamberini](https://discourse.cmake.org/user_avatar/discourse.cmake.org/vito.gamberini/32/4376_2.png) [@vito.gamberini](https://discourse.cmake.org/u/vito.gamberini)
#### Post date: [July 6, 2025, 8:53pm UTC](https://discourse.cmake.org/t/no-sources-given-to-target-cmake-cxx26-on-ubuntu-25-04/14240/17 "2025-07-06T20:53:50Z")

</div>

Whoever packaged the compiler.

If the packagers need support for their use case from the compiler, ex, they need to be able to tell the compiler “I’m going to put these files here” to make `-print-file-name` work, then it is on the packagers to communicate that the various compiler upstreams.

The compiler upstreams can decide on a case-by-case basis if they want to support the packagers conventions, or if it is on the packagers to align themselves with the compiler’s expectations.

CMake’s involvement begins and ends at `-print-file-name`. We 100% need better error messages if the path we get from `-print-file-name` is invalid, or the P3286 file contains bogus metadata, but we can’t “fix” that.

---

_[View the full topic](https://discourse.cmake.org/t/no-sources-given-to-target-cmake-cxx26-on-ubuntu-25-04/14240)._