I write plugins for a large, well known application.
Building the plugin is a complicated multi stage process with many dependencies.
I've spent a lot of time understanding the process (undocumented, only VS and Xcode project files are provided in the SDK).
I now have a single function call to create each plugin.
# project/meson.build
project_plugin_dep = declare_dependency(
include_directories: plugin_includes,
compile_args: plugin_c_args
)
# plugin/meson.build, repeated for each plugin
shared_module(
'plugin',
'plugin.c',
dependencies: project_plugin_dep
)
Without having seen what your user-defined function does, this is the best equivalent that I can come up with. If your plugins exist out of tree, then Meson can consume a pkg-config file. Postgres is slowly moving to Meson and uses out of tree plugins.There are a bunch of other steps. I have a YAML description of the plugin which I use to generate plugin skeleton, and so on.
Do you really think it makes sense to repeat that again and again?
And then, what if I need to fix a bug in the process?
In many large projects, the build system is necessarily large and complicated. It should be treated like any other code, and would massively benefit from the most basic tool for writing clean, readable code - user-defined functions.
Sadly, as much as Meson is tempting in other respects, without user-defined functions, it is unfit for human consumption.
The excuse Meson uses, that user-defined functions would make it Turing-complete and therefore not statically analyzable is totally false.
Adding user-defined functions without recursion does not make the language Turing-complete. On the contrary, as long as there is no recursion, you can inline the functions and create an equivalent code without functions.
Surely this is hyperbole since some of the largest open source projects consider it a viable build system. Systemd, Mesa, GStreamer, GNOME, Postgres.
Would I be correct in assuming that you are coming at this conversation having never tried Meson?
I tried to port the project I've previously mentioned here from CMake to Meson build, but could not find a way of doing it with Meson build in a satisfactory way.
The lack of user-defined functions is a show stopper because I have a complicated multis-stage build process. It is also very different between platform. Having to repeat parts of this process is disqualifying.
A programming language, and a build language is a programming language, without user-defined functions is unfit for human consumption.
The script can examine the dependencies, see if any of them changed and then rebuild only if needed.
It's like Greenspun's tenth law[1]:
> Any sufficiently complicated build script contains an ad hoc, informally-specified, bug-ridden, slow implementation of make.
But then, what do I need Meson for?
Meson provides custom targets, so Meson itself would examine the dependencies and rebuild if needed:
https://mesonbuild.com/Custom-build-targets.html
Obviously this might not do what you need specifically, but it does deal with some of the use cases for user-defined functions.
The entire point of having functions is to capture a "recipe" for building one type of target, and then be able to reuse it again and again.
Many other people like Meson, and many large projects have gone so far as to switch to using Meson. It's great that it worked for them, and it's OK that it didn't work for you.
That's exactly what I've been saying.
> As a result, you really dislike Meson, going so far as to call it "unfit for human consumption".
Meson got so many things right, but this one point makes it unusable for many projects, and prevents it from being industry standard.
Let me say it again, a programming language for writing large scale programs (11,000 lines is a large scale program) without functions/subroutines is unfit for human consumption.
It is made worse by giving a blatantly wrong rationale for not having user-defined function.
The only reason I even bothered spending so much time in responding is the hope that they will reverse this idiotic position.
Clearly there are people who think the position isn’t idiotic, and there are large projects which decided that Meson was a great choice and chose to migrate to it. This means that Meson is fit for human consumption.
The only thing I would use user-defined functions for is to structure the meson.build file in top-down manner, but each function would be called just once.
A lot of build systems won't support what you're asking here. Most don't do multi-stage builds, most don't have the concept of functions, many rely on external tools (autotools, pkg-config, etc). They're not "unfit for human consumption", which is pretty offensive to the people who chose Meson for their projects as well as Meson developers.
I get that you're offended by the situation. I might even agree with you in principle. But you're all over a Meson release thread, their 1.0 release thread even, shitting all over them, on Christmas Eve! Lighten up!
One thing that I find make CMake configurations completely unreadable is to try to have it do everything, coffee included. Where often a script could be run separately first (e.g. to generate those files), and then CMake could run normally afterwards.
No, it cannot be done before CMake runs. It is a target like any other build target. It depends on other files or targets, and should be rerun if any of them change, and if it is rebuilt, there are other targets that depend on it and should be rebuilt, and so on.
The entire point of the build system is to concisely and accurate specify the build graph, and only rebuild what is necessary.
For the big majority of projects out there, it seems to me that it is perfectly fine to handle that manually, because projects are not that big. Do we agree on that?
Then of course, if you are a GAFAM with a huge monorepo, probably you need more advanced tools (they don't build their own tools for nothing). My feeling, though, is that most people complaining that CMake is crap actually work on small projects and can't be bothered to run protoc manually "because their philosophy is that it should all work with one button".
Now it's not a pain for me, and apparently you're on the side of those who find cmake painful. Not sure how unreasonable my way is :)
Let's also agree that every time you change a .cpp file you compile it, and every time you change a .h file you compile all the affected .cpp files.
No we don't agree on that. That is the entire point of having a build system.
Even proto files you gave as an example, may import other proto files. When you change one of the imported files you have to recompile all proto files that import that file, transitively.
I have, unfortunately, worked with build systems that do not capture all dependencies and systems that capture way too many dependencies. The former resulted in many hours spent debugging non existent problems, and the latter resulted in many hours spent waiting for compilation to finish.
Most people either work on programs small enough where having a conservative approximation of the dependency graph is "fine" and excessive rebuilds are OK, or structure the project with a little repetition between the between the build time dependencies and the code. Or, they just let the module system in their language of language choice figure it out. These are all completely OK. But it's very expensive to do conservative recompilation and very error prone to do manual repetition in larger projects, and gets very complex in multi-language setups.
Where does the presupposition of just some build tools, but not the others (if I am reading it correctly), resulting in excessive rebuilds come from?
Apart from building a dependencies DAG for a project, the next fundamental responsibility of a build tool is to watch for / detect changed graph nodes and apply relevant build rules (expressed as graph edges) to rebuild only the necessary parts of the graph by traversing the path upward (if the apex node is up at the top). Naturally, if it is a dependency at the very bottom of a very leafy graph that has changed and upstream nodes have a dependency on it, it will trigger a large (of a very large) rebuild of the entire dependency path. But that is true of all build systems, not just select ones, and that is what we need buid tools for.
The problem is that many people either can't or don't express dependency rules correctly or dependencies generate build byproducts that are to be dynamically injected into the DAG, and they not properly accounted for in the build rules, and that results in an incorrectly built DAG at the build system run time resulting in excessive rebuilds.
In the ideal world, dependency management is relegated to a module system of a programming language. But since long serving incumbents cough cough C, C++ and Java (not until Java 9 anyway) do not have a module system, people have created a kludge (make) and later multiple kludges (autoconf/automake, cmake, maven/ivy etc) to bandaid the problem.
However a generic build tool can't account for a permutation of all possible scenarios, all possible programming languages, linkers, pre- and post-processors and whatever else needs to be done to produce a build artefact and excel at it in every possible use case. Some do it better than the others in very specific scenarios but do poorly in some other or fringe use cases. Then there are personal preferences. Therefore, the discourse over which build system is better than the other(s) will never end.
The module system is tasked with figuring out dependencies (i.e. files comprising a module), which is true for all programming languages starting from Standard ML through to Rust and anything in between.
Now it really feels that most people complaining about CMake and using Meson instead are not in this situation. And if they were, it would seem more pragmatic to me to say "CMake is fine for small projects, but at some size we needed something else" than "Meson is obviously, definitely orders of magnitude better because it took me less time learning how to compile my 20-files project".
And this isn't something that ultra-massive billion line projects singularly suffer from. You can easily suffer from all these issues and more in small and big software projects alike. I've suffered from these problems plenty in FOSS and commercial software that were small and large. "Computer program that generates another computer program" is actually a pretty normal and fine way to build software. But when compilation is expensive, these things matter, and you can't keep making excuses about it.
A build system should accurately track dependencies and rebuild things. It's that simple. You can say that running two commands is OK or whatever, but don't try to imply that people who want that criteria -- accurate deps and rebuilding -- aren't trying to solve problems they actually have, and then ditch the example when it becomes inconvenient to you that they are.
I don't try to have cmake do too much, e.g. I don't run the generation (or python scripts or custom functions) from cmake, because it quickly becomes a mess and running the steps manually / in a script is usually easy.
I haven't seen a project yet where the cmake config is not a mess (because let's be honest, most CMakeLists out there are pretty bad) and where this is clearly a limitation of cmake (and not of meson).
Happy to get an actual open source example where that's clearly the case, though.
and to me, build procedures that aren't completely implemented in cmake but require to install who knows what shell / platform / language / etc... in addition are the bane of my life as they make everything harder and more brittle and less cross-platform.
I don't get it. Also shell scripts are pretty good at running a couple commands, if really that's important.
I actually have maintained FLOSS projects, and in my experience, throwing everything in one cmake project makes it harder for contributors to understand. Because they only ever use one command, and the day something fails they have to dig into it to understand what that custom command does. On the contrary, if they need to run the generation script first, then they know that there is a generation step involved, and suddenly the whole thing feels less like magic.
So that is my opinion: if your FLOSS project uses a big custom build system (because you customized your cmake to do all sorts of custom things), then probably I will not contribute. Because I have other things to do than learning your custom integration of standards steps I actually would already know.
What's better is what is easier to understand and maintain. If your whole project is made of one file and builds with one simple call to a compiler, IMO it's fine to just have this line in the README, no need for a custom_target in cmake.
Now the thing is that often, I see that people put a lot of importance into the "it should be easy to get started for beginners who don't know the tools", and everything is glued together using some custom functions. The goal being that a beginner can essentially run "make" (yep, I've seen Makefiles wrapping CMakeLists for that) and the project will build (sometimes it will even install dependencies on the system, that's crazy). But then if the beginner wants to understand what's happening under the hood, well... it's a big pile of custom glue.
Now if, instead of that, you document the tools you are using in your README, and use the tools in a standard way, then people who know the tools will get started quickly ("oh, it's protoc then cmake, I know exactly how to use that"). And yeah, beginners will need to learn how to run those tools. But that's fine IMO, because if you use the tools the standard way (without custom functions), then the beginner can just look for the official docs.
Standard is better than custom.
There is definitely a religious feeling to some of their arguments.
My experience with QEMU's transition to meson is that it was clear from the beginning that Meson would not be able to do everything we needed, but it could do most of it. We live with the limitations and the outcome was extremely positive:
* 4000 less lines of build system gunk
* A lot fewer requests for help
* Several new features that could be implemented much more easily
* A transition that was
Of course there is still complexity, QEMU has 11000 lines of build system code. It was not easy and it involved several contributions to meson itself (that everybody now benefits from, for example Postgres) but it was totally worth it.
But, you're describing a 11,000 line program without the most basic abstraction mechanism known to man - user defined functions.
What are the benefits of not having user defined functions? none, zilch, nada.
The Meson team is grasping at straws trying to claim that having user-defined functions would make it Turing-complete. Without recursion, it won't.
Try to imagine how you would structure those 11,000 lines of code if user-defined functions were available? Would you really write it without extensive use of functions?
The only somewhat repetitive idiom is
foo = dependency('name', method: 'pkg-config', required: get_option('opt').disable_auto_if(condition))
to find a library, but otherwise there's very very little repetition of the kind that can be fixed with user-defined functions.
That's not surprising though. Because the user input is very constrained, I find it more natural to think of the build as "first prepare all these things, then use them to prepare all these other things" and so on, not as a complex program. As a very simple example, instead of abstracting "find flags for the C/C++ compiler" in a function and calling it twice, I prepare an array of languages and iterate on it.
It seems limiting but really it's just different and this kind of data flow mindset makes it very clear what is happening when (e.g. what files or arrays are read and which are written), and it also does not need a lot of abstractions (functions).
What it needs is decent data structures, and Meson's DSL is superior to CMake in that respect. So basically the code you write matches the tools you have.
Now I am not saying that Meson is perfect. But the functions thing is kind of a red herring. My #1 complaint is lack of include files (apart from going down to a subdirectory), not functions. #2 is messy handling of targets that generate files in subdirectories and #3 is a silly handling of some options.
The important thing is that none of these things, and not even functions for that matter, are baked in the language. There is room for fixing them in the future, because they are missing features or bugs rather than fundamental mismatches.
To be honest I believe that Meson should have used Starlark as the language, but it did not exist 10 years ago and anyway there is no Python implementation of it.
What thing you cannot do? I have used capnproto, generated c++ files and passed the result to the input of targets perfectly ok.
What I want to do is to spend the time and effort in understanding and debugging the process, and then encapsulating it in a single function call. Then completely forget about the details and use that single function call.
What is the advantage in not having user-defined functions? None, zilch, nada.
God awful ugly abomination of a language, but it does the work, and is almost universally supported.
I kinda get the rationale that no abstraction primitive should drive common patterns into the standard Meson distribution (an approach which has problems too, but at least it makes sense to me).
However, seeing function definitions rejected on the basis of Turing-completeness raised serious doubts about the project at the time, and things haven't changed much on this front apparently[1]
[1]: https://mesonbuild.com/Syntax.html#userdefined-functions-and...
I think there are a lot of exaggerations around here. I have been a Meson user for long moving from CMake. CMake is quite more frustrating.
Nothing, when applicable this works and we're doing it already, thank you.
But it doesn't as soon as one needs to abstract over CMake commands themselves (like add this target, set that target property, add this test and so on). Of course we could try to generate CMake files in wrapper scripts, but this is another level of CMake juggling entirely.
> CMake is quite more frustrating.
Could not agree more.
But even more frustrating than CMake programming is Meson checking almost all the boxes yet refusing to provide that last one bit of functionality that would make me switch to it (and giving a wrong justification for it).
Take waf for example: it’s effectively a Python framework for creating a custom build system, with every single internal feature accessible (and documented) with good design instead of abstractions. In waf, you can do things like:
* define custom build Tasks
* have those Tasks spawn other Tasks dynamically (e.g. by parsing the stdout of a compiler to find output files)
* Extend/modify the behavior of existing Task classes (e.g. the C/C++ linker)
* Modify the task scheduler to control execution order
* Implement commands that analyze your build scripts (e.g. to generate a compile_commands.json file, or anything else)
* cache any serializable Python types between build invications
* create reusable project-agnostic build scripts
* wrap everything in a nice API so that build scripts are manageable
* rich debugging features
Not to mention the expressive syntax of Python and the full power of the standard library.
And all you need is a Python interpreter (2.7 and 3+ both work), as Waf has zero dependencies.
The learning curve is high, but it’s worth it for projects that have a lot of build complexity. In my case, I use CMake until a project needs custom functions, or needs to support multiple languages, then I switch to waf.
In waf, it's possible to implement everything you need for that in a highly efficient way: the 'scanner' method that parses XML files to find linked resources to setup dependencies, the task that generates the header file with optional decompression API, etc. Because the task scheduler is flexible and programmable, you can implement it so that nothing runs unless it absolutely has to. Even the dependency scanner can cache data between builds so that it doesn't need to re-scan for dependencies unless any of the files it found in a previous invocation changed (hash, timestamp, etc)
For example: MyData.xml links to CoolStuff.xml; user modifies the whitespace of CoolStuff.xml, so MyData.xml needs to be rebuilt; the Task that converts it into a new representation runs, but since whitespace didn't affect the output of that task, the output file's hash is identical to what it was before, so the other tasks don't detect a change, and the build ends immediately.
In CMake, it would be impossible to implement something comparable. With custom functions/external scripts you can technically do anything, but there's a lot of inefficiency both in runtime and development effort. Since CMake doesn't actually perform builds itself, it's limited in what you can directly do, and the DSL isn't powerful enough to take advantage of more flexible CMake internals even if they existed.
But if you only need to compile C and C++ files with some light scripting, CMake is far superior for a lot of reasons.
At the end of the day, I don't want to have to know any build system let alone CMake. Period. Full stop.
I want my build system to be declarative. That way, when it breaks, and it will break, I can fix the dumb thing. (Here's an example of the kind of breakage that CMake has: I can't compile with "zig cc" and friends, because there's a space in that compile command and CMake has baked-in parsing. Whoops. Yeah, CMake is slowly fixing that, but it's going to take forever to roll out a fix for something that shouldn't be in scope of the build system.)
The fact that your project's build system needs complex features should be treated as a gigantic red klaxon to me, as a user, to stay far, far away from it. Needing to use complex features for build means that you need to reorganize your project and/or your "build system" needs to be broken apart into phases that are containing the complexity.
I am picking on CMake here, but this is a failure of any build system that tries to do "too much". Cargo, from Rust, has similar problems.
While backporting a new PipeWire release to my linux distro earlier this year, I discovered that it had added dependencies on Meson features so new that they didn't exist in Debian Stable. User-defined functions would have allowed me to shim those features for the older Meson version, greatly simplifying my work and minimizing my patches' footprint.
I'm not sure if that situation is common enough to justify user-defined functions to Meson's author, but it was frustrating, and it seemed ironic that a build system designed to smooth over platform compatibility problems had itself become one.
The main culprit is GNOME packages because they were early adopters of Meson and thus had a few incorrect idioms that were cargo culted to a few dozen packages. For a similar issue in RHEL we decided to update Meson, but we needed to revert some changes that tightened the language. I do plan to include those changes in Meson in the near future, keyed on the minimum version in the "project" line, so that older packages are easier to build with newer Meson.
That said, did you consider backporting the new features to the older Meson in Debian stable?
No, since modifying an additional project would run counter to the goal of minimizing the scope of the backport's changes. (Especially important when said project is a dependency of other packages, and could potentially break other things on the distro. This was Debian Stable, after all.)
It does not. This backport was not for a single user, but for deployment into the package repo used by multiple systems (including build systems). A one-off Meson build would be unsuitable for this case.
In any case, note that I didn't claim it was impossible without user-defined functions. It was simply messier and more painful, much as a special Meson build would be.
And they definitely don't make a language Turing-complete.
They just make large code bases tractable for humans.
Turing-completeness is, itself, often a red herring. What it alludes to is the difficulty in predicting what a program would do because, in the general case, it is impossible to tell what behaviour a program in a Turing-complete language would exhibit for a particular input short of observing what it actually does. But even languages that are not Turing-complete often fall far beyond what is practically predictable. While theoretically decidable, predicting what a program written in a language with nothing but boolean variables and loops of no more than two iterations, or, alternatively, functions of boolean parameters even without recursion, is PSPACE-complete (as such a language can easily encode TQBF [1]). I.e. predicting how such a program would behave is computationally intractable, since it grows exponentially with the size of the program. Programs in such a language would also fall outside the capabilities of heuristic methods, such as SAT solvers, to tackle (SAT is merely NP-hard compared to the strictly harder PSPACE-hard TQBF).
So if your language has variables and loops (even limited to two iterations), or variables and functions (even without recursion) -- it's intractable, even though it's not Turing-complete. I don't know whether this particular language qualifies or not.
[1]: https://en.wikipedia.org/wiki/True_quantified_Boolean_formul...
The entire issue is a red herring.
They climbed up a tree and can't climb down without losing face.
The correct way to handle the situation is to make the change, wait for everyone to cool down, and then pretend that was their position all along.
> Meson does not currently support user-defined functions or methods. The addition of user-defined functions would make Meson Turing-complete which would make it harder to reason about and more difficult to integrate with tools like IDEs.
No, user-defined functions won't make Meson Turing-complete as long as recursion is not allowed.
User defined functions make it easier to encapsulate complicated build logic and reuse it.