Document 'meson subprojects update --reset'

[andy5995]

tree-sitter wasn't shown in the output but is on L79 of meson.build.

vs-templ| Project name: vs-templ
vs-templ| Project version: 0.3.1
vs-templ| C++ compiler for the host machine: clang++-19 (clang 19.1.3 "Debian clang version 19.1.3 (2)")
vs-templ| C++ linker for the host machine: clang++-19 ld.bfd 2.43.1
vs-templ| Program pandoc found: NO
vs-templ| Build targets in project: 53
vs-templ| Subproject vs-templ finished.


meson.build:79:18: ERROR: Subproject exists but has no meson.build file.

Removing subprojects/tree-sitter and re-running bun run meson-setup.clang-release fixed the issue.

Looking briefly at the tree-sitter wrap, the problem may be with using the subproject dir as the patch directory. Paradox?

[KaruroChori]

Looking briefly at the tree-sitter wrap, the problem may be with using the subproject dir as the patch directory. Paradox?

My understanding is that writing patch_directory = tree-sitter uses ./subprojects/packagefiles/tree-sitter as folder from which meson.build is copied. That is what I understood by reading the docs at least. Is that wrong?

For ref: https://mesonbuild.com/External-Project-module.html#using-wrap-file

I would have assumed meson patches file back after updating its git deps. If not, I think it is an upstream issue.

[andy5995]

I believe you are correct. I just looked through the structure of the wrapdb project and it matches what you state.

I opened an issue upstream

[andy5995]

See mesonbuild/meson#13970 (comment) for the fix.

I'll re-title this issue to suggest documenting it here.

Note I didn't read through the meson docs or the dupe issue to see if there are cases when the reset arg should not be appended.

[KaruroChori]

Thanks for figuring this out. To be honest I feels like meson is doing the wrong thing in this case.
If there had been a --patch to be applied both when first install and for each update, I would have considered it consistently annoying.

But the fact updating a git wrap does not apply the patch automatically after the first time makes no sense in my head :/.
Patches are almost always added because the underlying repo would not be able to work with meson, so why would someone not want them applied as a default.

In any case, I will add a note in the build docs.

[andy5995]

But the fact updating a git wrap does not apply the patch automatically after the first time makes no sense in my head :/. Patches are almost always added because the underlying repo would not be able to work with meson, so why would someone not want them applied as a default.

Confusing to me as well. If the patch is always wiped when updating, that just basically means you always have to use --reset, unless you don't have any subprojects that uses patches.