After exactly 10 years, Meson 1.0.0 is out

The first ever commit to the Meson repository was made 10 years ago to this day. To celebrate we have just released the long-awaited version 1.0.

The original design criterion for doing a 1.0 release was "when Meson does everything GStreamer needs". This happened, checks notes, three years ago (arguably even earlier than that). That is not the world's fastest reaction time, but that comes mostly down to our development policy. Meson aims to make releases at a steady pace and maintains backwards compatibility fairly well (not perfectly). There is unlikely to ever be a big breaking change, so there is no pressing technical need to bump the major version number.

Thus 1.0 is mostly a symbolical milestone rather than a technical one, end users should not not really notice that big of a difference. This does not mean that the release is any less important, though. To celebrate here is an assortment of random things that have happened over the years. Enjoy.

The greatest achievement

Perhaps the best example demonstrating the maturity of Meson is that I no longer do all the decisions. In fact most decisions and especially the code that goes with it is done by a diverse group of people. In fact I do very little of the actual development, I'm more of a product owner of sorts that can only nudge the project into certain directions rather than being able to turn the entire ship around on a dime. This is a bit sad, but absolutely necessary for long term survival of the project. It means that if one of those malevolent buses that seem to stalk software developers succeeded in hitting me, its effect on the project would not be all that big.

Reimplementation

There are two main reasons for reimplementing an existing open source project from scratch. The first one is that the upstream developer is a jerk and people don't want to work with them. The second is that someone, somewhere sees the project as important enough to have a second, independent implementation. I'm happy to report that (as far as I know at least), Meson is in the second group because there is a second from-scratch implementation of Meson called Muon. 

Meson is implemented in Python but its design was from the very start that this is only an implementation detail. We spent a fair bit of effort ensuring that the Python bits don't leak in the DSL, even by accident. There wasn't really any way of being sure about it short of doing a second implementation and now there is one as Muon is implemented in plain C.

Collaborative design

We all know that disagreeing with other people on the Internet might be very discouraging. However sometimes it works out incredibly well, such as in this merge request. That MR was really the first time a new feature was proposed and the submitter had a very different idea than me of what the API should look like. I distinctly remember feeling anxious about that at the time because I basically had to tell the submitter that their work would not be accepted.

To my surprise everything went really well. Even though there were many people involved and they had wildly different ideas on how to get the feature done, there was no pouting, no stomping of feet, shouting or the like (which, for the record, there had been in other similar discussions). Absolutely everybody involved really wanted to get the feature in and were willing to listen to others and change their stances based on the discussion. The final API turned out to be better than any of the individual proposals.

Thanks to contributors

According to Github statistics, a total of of 717 different people have at least one commit in the repository. This number does not cover all the other people who have contributed in other ways like docs, bug triaging, converting existing projects and so on. It is customary to thank people who have done major contributions like new features in milestone posts like these.

I'm going to do something different instead. In addition to "the big stuff" any project has a ton of less than glamorous work like bug fixing, refactoring, code cleanups and the like. These tasks are just as important as the more glitzy ones, but it sadly go underappreciated in many organisations. To curb this trend I'd like to pick three people to thank for the simple reason that when it turned out that sh*t needed to be done, they rolled up their sleeves and did it. Over and over again.

The first one is Dylan Baker, who has done major reorganisation work in the code including adding a lot of typing hints and fixed the myriad of bugs the adding of said type hints uncovered.

The second person is Eli Schwartz, who has done a ton of work all around, commented on many bug reports and on the Matrix channel. In fact he has done so much stuff that I suspect he never sleeps.

And finally we have Rosen Penev, who has done an astounding amount of work on WrapDB, both fixing existing wraps as well as updating them to new releases.

And finally: a secret

Meson gets a lot of bug reports. A lot a lot. Nirbheek Chauhan, one of the co-maintainers, once told me that Meson generates more bug email than all Gnome projects combined. I try my best to keep up with them, but the sad truth is that I don't have time to read most of them. Upon every release I have to clean up my mailbox by deleting all Meson bug mail.

The last time I did this I nuked more than 500 email threads in one go. No, not emails, email threads. So if you have wondered why your bug report has not gotten any replies, this is why. Simply reading the contents of Meson bug emails would be more than a full time job. Such is the price of success, I guess.

Print quality PDF generation, color separations, other fun stuff

Going from the simple color managed PDF generator discussed in the previous blog post into something more useful requires getting practical. So here is a screenshot of a "print ready" PDF document I generated with the code showing a typical layout you'd use for a softcover book. As printers can't really print all the way to the edges of paper, the cover needs to be printed to a larger sheet and then cut to its final size.

It's not of artistically high quality, granted, but most of the technical bits are there:

• The printed page is noticeably bigger than the "active area" and has a bunch of magic squiggles needed by the printing house
• The output is fully color managed CMYK
• The gray box represents the bleed area and in a real document the cover image would extend over it, but I left it like this for visualization purposes.
• Text can be rendered and rotated (see spine)
• The book title is rendered with gold ink, not CMYK inks
• There are colorbars for quality control
• The registration and cut marks (the "bullseyes" and straight lines at paper corners, respectively) are drawn on all plates using PDF's builtin functionality so they are guaranteed to be correctly aligned
• None of the prepress marks are guaranteed to be actually correct, I just swiped them from various sources

The full PDF can be downloaded from this link. From this print PDF, we can generate separations (or "printing plates") for individual ink components using Ghostscript.

Looking at this you can find several interesting things. For example the gray box showing the bleed area is composed of C, M and Y inks instead of only K, even though it was originally defined as a pure gray in RGB. This is how LittleCMS chose to convert it and it might or might not be what the original artist had in mind. High quality PDF generation is full of little quirks like this, blindly throwing numbers at color conversion functions is not enough to get good results, end users might need fairly precise control over low level operations.

Another thing to note is how the renderer has left "holes" for the book title in CMYK plates even though all color is in the gold ink plate. This avoids mixing inks but on the other hand requires someone to do proper trapping. That is its own can of worms, but fortunately most people can let the RIP handle it (I think).

Color management, this time with PDF

In previous posts the topic of color management in Cairo was examined. Since then people have told me a few things about the issue. According to them (and who am I do to proper background research and fact checking, I'm just someone writing on the Internet) there are a few fundamental problems with Cairo. The main one is that Cairo's imaging model is difficult to implement in GPU shaders. It also is (again, according to Internet rumors) pretty much impossible to make work with wide gamut and HDR content.

Dealing with all that and printing (which is what I was originally interested in) seems like a too large a mouthful to swallow. One thing lead to another and thus in the spirit of Bender, I wrote my own color managed PDF generator library. It does not try to do any imaging or such, just exposes the functionality that is in the PDF image and document model directly. This turned out to take surprisingly little work because this is a serialization/deserialization problem rather than an image processing one. You just dump the draw commands and pixels to a file and let the PDF viewer take care of showing them. Within a few days I had this:

This is a CMYK PDF that is fully color managed. The image on the second page was originally an RGB PNG image with an alpha channel  that was converted to CMYK automatically. The red square is not part of the image, it is there to demonstrate that transparency compositing works. All drawing commands use the /DeviceCMYK color space. When creating the PDF you can select whether the output should be in RGB, grayscale or CMYK and the library automatically generates the corresponding PDF commands. All of the heavy lifting is done by LittleCMS, there are no unmanaged color conversions in the code base.

Since everything was so straightforward, I went even further. That screenshow is not actually showing a CMYK PDF. The yellow text on the purple background is a spot color that uses a special gold ink. Thus the PDF has five color channels instead of four. Those are typically used only in high quality print shops for special cases like printing with specific Pantone inks or specifying which parts of the print should be em/debossed, varnished or the like.

What would it take to use this for realsies?

There does seem to be some sort of a need for a library that produces color managed PDFs. It could be used at least by Inkscape and Gimp, possibly others as well. In theory Cairo could also use it for PDF generation so it could delete its own PDF backend code (and possibly also the PostScript one) and concentrate only on pushing and compositing pixels. Then again, maybe its backwards compatibility requirements are too strict for that.

In any case the work needed splits neatly into two parts. The first one is exposing the remaining functionality in PDF in the API. Most of it is adding functions like "draw a bezier with values ..." and writing out the equivalent PDF command. As the library itself does not even try to have its own imaging model, it just passes things directly on. This takes elbow grease, but is fairly simple.

The other part is text. PDF's text model predates Unicode so it is interesting to say the least. The current code only supports (a subset of) PDF builtin fonts and really only ASCII. To make things actually work you'd probably need to reimplement Cairo's glyph drawing API. Basically you should be able to take PangoCairo, change it a bit and point it to the new library and have things work just as before. I have not looked into how much work that would actually be.

There are currently zero tests and all validation has been done with the tried and true method of "try the output on different programs and fix issues until they stop complaining". For real testing you'd need access to a professional level printer or Adobe Acrobat Pro and I have neither.

Going inside Cairo to add color management

Before going further you might want to read the previous blog post. Also, this:

I don't really have prior experience with color management, Cairo internals or the like. I did not even look at the existing patchsets for this. They are fairly old so they might have bitrotted and debugging that is not particularly fun. This is more of a "the fun is in the doing" kind of thing. What follows is just a description of things tried, I don't know if any of it would be feasible for real world use.

Basically this is an experiment. Sometimes experiments fail. That is totally fine.

Main goals

There are two things that I personally care about: creating fully color managed PDFs (in grayscale and CMYK) and making the image backend support images in colorspaces other than sRGB (or, more specifically, "uncalibrated RGB which most of the time is sRGB but sometimes isn't"). The first of these two is simpler as you don't need to actually do any graphics manipulations, just specify and serialize the color data out to the PDF file. Rendering it is the PDF viewer's job. So that's what we are going to focus on.

Color specification

Colors in Cairo are specified with the following struct:

struct _cairo_color {

    double red;

    double green;

    double blue;

    double alpha;

    unsigned short red_short;

    unsigned short green_short;

    unsigned short blue_short;

    unsigned short alpha_short;

};

As you can probably tell it is tied very tightly to the fact that internally everything works with (uncalibrated) RGB, the latter four elements are used for premultiplied alpha computations. It also has a depressingly amusing comment above it:

Fortunately this struct is never exposed to the end users. If it were it could never be changed without breaking backwards compatibility. Somehow we need to change this so that it supports other color models. This struct is used a lot throughout the code base so changing it has the potential to break many things. The minimally invasive change I could come up with was the following:

struct _comac_color {

    comac_colorspace_t colorspace;

    union {

        struct _comac_rgb_color rgb;

        struct _comac_gray_color gray;

        struct _comac_cmyk_color cmyk;

    } c;

};

The definition of rbg color is the original color struct. With this change every usage of this struct inside the code base becomes a compile error which is is exactly what we want. At every point the code is changed so that it first asserts that colorspace is RGB and then accesses the rgb part of the union. In theory this change should be a no-op and unless someone has done memcpy/memset magic, it is also a no-op in practice. After some debugging (surprisingly little, in fact) this change seemed to work just fine.

Color conversion

Ideally you'd want to use LittleCMS but making it a hard dependency seems a bit suspicious. There are also use cases where people would like to use other color management engines and even select between them at run time. So a callback functions it is:

typedef void (*comac_color_convert_cb) (comac_colorspace_t,
                                        const double *,
                                        comac_colorspace_t,
                                        double *,
                                        comac_rendering_intent_t,
                                        void *);

This only converts a single color element. A better version would probably need to take a count so it could do batch operations. I had to put this in the surface struct, since they are standalone objects that can be manipulated without a drawing context.

Generating a CMYK PDF stream is actually fairly simple for solid colors. There are only two color setting operators, one for stroking and one for nonstroking color (there may be others in e.g. gradient batch definitions, I did not do an exhaustive search). That code needs to be changed to convert the color to match the format of the output PDF and then serialize it out.

CMYK PDF output

With these changes creating simple CMYK PDF files becomes fairly straightforward. All you need to do as the end user is to specify color management details on surface creation:

comac_pdf_surface_create2 (

    "cmyktest.pdf",

    COMAC_COLORSPACE_CMYK,

    COMAC_RENDERING_INTENT_RELATIVE_COLORIMETRIC,

    comac_default_color_convert_func,

    NULL,

    595.28,

    841.89);

and then enjoy the CMYK goodness:

What next?

Probably handling images with ICC color profiles.

Experimenting on how to add CMYK and color management to Cairo

Cairo is an amazing piece of tech that powers a lot of stuff, like all of GTK. Unfortunately it is not without its problems. The biggest one being that it was designed almost 20 years ago with the main use case of dealing with "good old" 8 bit uncalibrated RGB images. There has been a lot of interest in adding native support for things like CMYK documents, linear RGB, color calibration, wide gamuts and all of that good stuff. Sadly it has not come to be.

The reasons are mostly the same as always. The project is sadly understaffed and there does not seem to be a corporate sponsor to really drive the development forward. What makes things extra difficult is that Cairo supports a lot of different platforms like Postscript, Win32, Quartz and SVG. So if someone wants to add new features in Cairo, not only do they need to understand how color math works and how to do C, they would also need to handle all the various backends. That is a rare combination of skills. In any case the patchset needed to make all that happen would be enormous and thus hard to get reviewed and merged.

As an exercise I thought I'd see if I could change the landscape somewhat to make it easier to experiment with the code base. Out of this came this project repo called Comac (for, obviously, COlor MAnaged Cairo). The creation process was fairly straightforward:

• Take the latest Cairo trunk
• Delete every backend except image and PDF
• Rename all files and symbols from cairo-something to comac-something
• Add minimal code for generating grayscale and CMYK PDFs
• Create a demo app that creates test PDFs

The output files are very simple, but the whole thing actually works. Both Okular and Acrobat Reader will happily display the documents without errors. Thus people who are interested in color work can now actually look into it without having to understand how Xlib works. Since Comac does not need to follow all of Cairo's backwards compatibility guarantees, experimenters can play a bit more fast & loose.

Want to contribute?

Merge requests welcome, I guess? I have not thought too deeply about governance and all that. If there is a lot of interest, this could be moved to its own top level project rather than living in my Github pages. Just note that the goal is not to fork things and create yet another graphics library. In the best possible case this project is only used to discover a good API and an idea on how it could best be implemented. These changes could then be put into Cairo core.

If you don't tolerate it in new code you should not tolerate it in old code either

Let's assume that you are working on a code base and notice that it has some minor issue. For argument's sake we'll say that it has some self written functionality and that the language's standard library has added identical functionality recently. Let's further assume that that said implementation behaves exactly the same as the self written one. At this point you might decide to clean up the code base, make it use the stdlib implementation and delete the custom code. This seems like a nice cleanup so you then file merge request to get the thing changed.

It might be accepted and merged without problems. On the other hand it might be blocked, either temporarily or permanently. There are several valid reasons for not merging the change. For example:

1. That part of the code does not have sufficient tests so we don't know if the change is safe.
2. A release is coming up soon, so we are minimizing all changes that might cause regressions, no matter how minor. The merge can only be done after the release is out.
3. We need to support old platform X, whose stdlib does not have that functionality.
4. It's not immediately obvious that the two implementations behave identically (for example, because the old implementation has load-bearing bugs) so a change might introduce bugs.

Getting blocked like this is a bit unfortunate, but these things happen. The important thing, however, is that all these are solid technical reason for not doing the cleanup (or at least not do it immediately). Things get bad when you get blocked by some other reason, such by your reviewer asking "why are you even doing this at all". This is a reasonable question, so let's examine it in detail.

Suppose that instead of submitting a cleanup commit you are instead submitting a piece of completely new functionality. In this MR you have chosen to reimplement a piece of standard library code (for no actual gain, just because you were not aware of its existence). The review comment that you should get is "You are reimplementing stdlib functionality, delete that code here and use the standard library instead". This is a valid review comment and something that should be heeded.

The weird thing here is that this is in a way the exact same change, but it is either not acceptable or absolutely necessary depending on whether parts of the code are already inside your repo or not. This is weird and should not be the case, but human beings are strangely irrational and their "value functions" are highly asymmetric. This can lead to lots of review fighting if one person really wants to fix the issue whereas some other one does not see the value in it. The only real solution is to have a policy on this, specifically that submitting a change that fixes an issue that would be unacceptable in new code is, by itself, a sufficient reason to do the work but not to merge it without technical review. This shifts the discussion from "should this be done at all" to "what are the technical risks and merits of this change", which is the way reviews should be done.

I should emphasize that this does not mean that you should go and immediately spend 100% of your time fixing all these existing issues in your code base. You could, but probably it is not worth it. What this guideline merely says that if you happen to come across an issue like this and if you feel like fixing it then you can do it and reviewers can't block you merely by "I personally don't like this" line of reasoning.

A nastier version

Suppose again you are a person who cares about the quality of your work. That you want to write code that is of sufficiently good quality and actually care about things like usability, code understandability and long term maintainability. For you people there exists a blocker comment that is much worse than the one above. In fact is the single biggest red flag for working conditions I have ever encountered:

But we already have [functionality X] and it works.

This question does have a grain of truth in it, existing code does have value. Sadly it is most commonly used as a convenient way to hardblock the change without needing to actually think about it or having to explain your reasoning.

Several times I've had to make some change into existing code and hours and days of debugging later found that the existing self written code has several bugs that a stdlib implementation definitely would not have. After reporting my findings and suggesting a cleanup the proposal has been knocked out with reasoning above. As a consultant I have to remain calm and respect the client's decision but the answer I have wanted to shout every time is: "No it's not! It's completely broken. I just spent [a large amount of time] fixing it because it was buggy and looking at the code makes me suspect that there are a ton of similar bugs in it. Your entire premise is wrong and thus every conclusion drawn from it is incorrect."

And, to reiterate, even this is not a reason by itself to actually go and change the code. It might get changed, it might not. The point is the prevailing attitude around minor fixups and how it matches your personal desire of getting things done. If you are a "fixing things proactively makes sense" kind of person you are probably not going to be happy in a "let's hide under the bed and pretend everything is fine" kind of organization and vice versa.

Making Visual Studio compilers directly runnable from any shell (yes, even plain cmd.exe)

The Visual Studio compiler toolchain behaves in peculiar ways One of the weirdest is that you can't run the compiler from any shell. Instead you have to run the compiler either from a special, blessed shell that comes with VS (the most common are "x86 native tools shell" and "x64 native tools shell", there are also ARM shells as well as cross compilation shells) or by running a special bat file inside a pristine shell that sets things up. A commonly held misbelief is that using the VS compiler only requires setting PATH correctly. That is not true, it requires a bunch of other stuff as well (I'm not sure if all of that is even documented).

To anyone who has used unixy toolchains, this is maddening. The classic Unix approach is to have compiler binaries with unique names like a hypothetical armhf-linux-gcc-11 from any shell. Sadly this VS setup has been the status quo for decades now and it is unlikely to change. In fact, some times ago I had a discussion with a person from Microsoft where I told them about this problem and the response I got back was, effectively: "I don't understand what the problem is" followed by "just run the compiles from the correct shell".

So why is this a bad state of things then? There are two major issues. The first one is that you have to remember how every one of your build trees has been set up. If you accidentally run a compilation command using the wrong shell, the outcome is very undefined. This is the sort of things that happens all the time because human beings are terrible at remembering the states of complicated systems and specific actions that need to be taken depending on their state (as opposed to computers, which are exceptionally good at those things). The second niggle is that you can't have two different compilers active in the same shell at the same time. So if, for example, you are cross compiling and you need to build and run a tool as part of that compilation (e.g. Protobuf) then you can't do that with the command line VS tools. Dunno if it can be with solution files either.

Rolling up them sleeves

The best possible solution would be for Microsoft to provide compiler binaries that are standalone and parallel runnable with unique names like cl-14-x64.exe. This seems unlikely to happen in the near future so the only remaining option is to create them ourselves. At first this might seem infeasible but the problem breaks neatly down into two pieces:

• Introspect all changes that the vsenv setup bat file performs on the system.
• Generate a simple executable that does the same setup and then invokes cl.exe with the same command line arguments as were given to it.

The code that implements this can be found in this repository. Most of it was swiped from Meson VS autoactivator. When you run the script in a VS dev tools shell (it needs access to VS to compile the exe) you get a cl-x64.exe that you can then use from any shell. Here we use it to compile itself for the second time:

Downsides

Process invocation on Windows is not particularly fast and with this approach every compiler invocation becomes two process invocations. I don't know enough about Windows to know whether one could avoid that with dlopen trickery or the like.

For actual use you'd probably need to generate these wrappers for VS linkers too.

You have to regenerate the wrapper binary every time VS updates (at least for major releases, not sure about minor ones).

The end results has not been tested apart from simple tests. It is a PoC after all.