It looks like they posted it:
@vito.gamberini It looks like one of the YouTube comments is complaining about how CMake FILE_SETs handle header-only libraries. Have you seen that before?
It looks like they posted it:
@vito.gamberini It looks like one of the YouTube comments is complaining about how CMake FILE_SETs handle header-only libraries. Have you seen that before?
Theyâre simply wrong. Not interesting wrong, not niche misunderstanding wrong, not even outdated knowledge wrong. Theyâre â2 + 2 = 5â wrong
Having flipped the talk on in âthe background yesterdayâ, I might be wrong â but I think only header file sets where shown on the slides (see minute 57 or so). Perhaps thatâs the confusion?
btw: Itâs scary that Iâd already forgotten 90% of the talk even having been present â I guess thatâs bc cmake stuff at âthe day jobâ is usually that routine adding files, etc.
The most generous interpretation is that they donât understand how install() interacts with FILE_SET. But CMake throws big âyou must say install(FILE_SET)" errors if you ever make the attempt.
CMake, like C++, is an expert tool. Thatâs maybe not a good thing, not something to be super proud of or whatever, but itâs true. Like any expert tool, there are a countless people who hold it wrong and then complain it doesnât work. Complaints should largely be interpreted through this lens. Someone who is complaining, âX doesnât workâ is usually saying âX is poorly communicated and arcaneâ.
I didnât cover C++ module file sets, no. It seemed premature for a talk entitled Pragmatic CMake to cover building and shipping C++ modules. I would expect more practical experience first.
Clearly, no experts commented on YouTube so far.