% language=uk \environment luametatex-style \startcomponent luametatex-introduction \startchapter[title=Introduction] Around 2005 we started the \LUATEX\ projects and it took about a decade to reach a state where we could consider the experiments to have reached a stable state. Pretty soon \LUATEX\ could be used in production, even if some of the interfaces evolved, but \CONTEXT\ was kept in sync so that was not really a problem. In 2018 the functionality was more or less frozen. Of course we might add some features in due time but nothing fundamental will change as we consider version 1.10 to be reasonable feature complete. Among the reasons is that this engine is now used outside \CONTEXT\ too which means that we cannot simply change much without affecting other macro packages. However, in reaching that state some decisions were delayed because they didn't go well with a current stable version. This is why at the 2018 \CONTEXT\ meeting those present agreed that we could move on with a follow up tagged \METATEX, a name we already had in mind for a while, but as \LUA\ is an important component, it got expanded to \LUAMETATEX. This follow up is a lightweight companion to \LUATEX\ that will be maintained alongside. More about the reasons for this follow up as well as the philosophy behind it can be found in the document(s) describing the development. During \LUATEX\ development I kept track of what happened in a series of documents, parts of which were published as articles in user group journals, but all are in the \CONTEXT\ distribution. I did the same with the development of \LUAMETATEX. The \LUAMETATEX\ engine is, as said, a follow up on \LUATEX. Just as we have \CONTEXT\ \MKII\ for \PDFTEX\ and \XETEX, we have \MKIV\ for \LUATEX. For \LUAMETATEX\ we have yet another version of \CONTEXT: \LMTX. By freezing \MKII, and at some point freezing \MKIV, we can move on as we like, but we try to remain downward compatible where possible, something that the user interface makes possible. Although \LUAMETATEX\ can be used for production we can also use it for possibly drastic experiments but without affecting \LUATEX. Because we can easily adapt \CONTEXT\ to support both, no other macro package will be harmed when (for instance) the interface that the engine provides change as part of an experiment or cleanup of code. Of course, when we consider something to be useful, it can be ported back to \LUATEX, but only when there are good reasons for doing so and when no compatibility issues are involved. By now the code of these two related engines differs a lot so in retrospect it makes less sense to waste time on backporting anyway. When considering this follow up one consideration was that a lean and mean version with an extension mechanism is a bit closer to original \TEX. Of course, because we also have new primitives, this is not entirely true. The basic algorithms remain the same but code got reshuffled and because we expose internals names of variables and such are sometimes changed, something that is noticeable in the token and node interfaces. Delegating tasks to \LUA\ already meant that some aspects, especially system dependent ones, no longer made sense and therefore had consequences for the interface at the system level. In \LUAMETATEX\ more got delegated, like all file related operations. The penalty of moving more responsibility to \LUA\ has been compensated by (hopefully) harmless optimization of code in the engine and some more core functionality. This manual started as an adaptation of the \LUATEX\ manual and therefore looks similar. Some chapters are removed, others were added and the rest has been (and will be further) adapted. It also discusses the (main) differences. Some of the new primitives or functions that show up in \LUAMETATEX\ might show up in \LUATEX\ at some point, but most will be exclusive to \LUAMETATEX, so don't take this manual as reference for \LUATEX ! As long as we're experimenting we can change things at will but as we keep \CONTEXT\ \LMTX\ synchronized users normally won't notice this. Often you can find examples of usage in \CONTEXT\ related documents and the source code so that serves a reference too. For \CONTEXT\ users the \LUAMETATEX\ engine will become the default. As mentioned, the \CONTEXT\ variant for this engine is tagged \LMTX. The pair can be used in production, just as with \LUATEX\ and \MKIV. In fact, most users will probably not really notice the difference. In some cases there will be a drop in performance, due to more work being delegated to \LUA, but on the average performance will be better, also due to some changes below the hood of the engine. Memory consumption is also less. The timeline of development is roughly: from 2018 upto 2020 engine development, 2019 upto 2021 the stepwise code split between \MKIV\ and \LMTX, while in 2020 we will (mostly) freeze \MKIV\ and \LMTX\ will be the default. As this follow up is closely related to \CONTEXT\ development, and because we expect stock \LUATEX\ to be used outside the \CONTEXT\ proper, there will be no special mailing list nor coverage (or pollution) on the \LUATEX\ related mailing lists. We have the \CONTEXT\ mailing lists for that. In due time the source code will be part of the regular \CONTEXT\ distribution so that is then also the reference implementation: if needed users can compile the binary themselves. This manual sometimes refers to \LUATEX, especially when we talk of features common to both engine, as well as to \LUAMETATEX, when it is more specific to the follow up. A substantial amount of time went into the transition and more will go in, so if you want to complain about \LUAMETATEX, don't bother me. Of course, if you really need professional support with these engines (or \TEX\ in general), you can always consider contacting the developers. % And yes, I'm really fed up with receiving mails or seeing comments where there's % this 'always need to be present' negative remark (nagging) about the program, % documentation, development, support, etc. present, probably to put the writer on % a higher stand, or maybe to compensate some other personal shortcoming ... who % knows. This 'i need to make my stupid point' behaviour seems to come with the % internet and it also seems to increase, but that doesn't mean that I want to deal % with those unpleasant people for the sake of the larger "tex good". Therefore, I'm % quite happy in the nearly always positive and motivating ConTeXt bubble. It's also % why I (start) avoid(ing) certain mailing lists and don't really follow forums. \blank[big] Hans Hagen \blank[2*big] \starttabulate[|||] \NC Version \EQ \currentdate \NC \NR \NC \LUAMETATEX \EQ \cldcontext{LUATEXENGINE} % \cldcontext{LUATEXVERSION} / % \cldcontext{LUATEXFUNCTIONALITY} \NC \NR \NC \CONTEXT \EQ MkIV \contextversion \NC \NR \NC \LUATEX\ Team \EQ Hans Hagen, Hartmut Henkel, Taco Hoekwater, Luigi Scarso \NC \NR \stoptabulate \vfilll {\bf remark:} \LUAMETATEX\ development is mostly done by Hans Hagen and in adapting the macros to the new features Wolfgang Schuster, who knows the code inside||out is a instrumental. In the initial phase Alan Braslau, who love playing with the three languages did extensive testing and compiled for several platforms. Later Mojca Miklavec make sure all compiles well on the buildbot infrastructure. After the first release more users got involved in testing. Many thanks for their patience! The development also triggered upgrading of the wiki support infrastructure where Taco Hoekwater and Paul Mazaitis have teamed up. So, progress all around. {\bf remark:} When there are non|-|intrusive features that also make sense in \LUATEX, these will be applied in the experimental branch first, so that there is no interference with the stable release. However, given that in the meantime the code bases differs a lot, it is unlikely that much will trickle back. This is no real problem as there's not much demand for that anyway. {\bf remark:} Most \CONTEXT\ users seem always willing to keep up with the latest versions which means that \LMTX\ is tested well. We can therefore safely claim that end of 2019 the code has become quite stable. There are no complaints about performance (on my 2013 laptop this manual compiles at 24.5 pps with \LMTX\ versus 20.7 pps for the \LUATEX\ manual with \MKIV). After updating some of the \CONTEXT\ code to use recently added features by the end of 2020 I could do more than 25.5 pps but don't expect spectacular bumps in performance (I need a new machine for that to happen). Probably no one notices it, but memory consumption stepwise got reduced too. And \unknown\ the binary is still below 3~MegaBytes on all platforms. \stopchapter \stopcomponent % I'm not that strict with incrementing numbers, but let's occasionally bump the % number. Once we're stable it might happen more systematically. For sure I don't % want to end up with these 0.99999999 kind of numbers. % We started with 2.00.0 and kept that number till November 2019, after Alan % Braslau and I did the initial beta release at April 1, 2019. After more than a % year working on the code base after the \CONTEXT\ 2019 meeting a state was % reached where nothing fundamental got added for a while. When \LUATEX\ needs a % patch, I check the \LUAMETATEX\ code base as the same patch might be needed % there. On the other hand, we don't need a strict compatibility, so some patched % in \LUATEX\ are not applied here. % % In November 2019 I started wondering if we should bump the number, just for the % sake of showing that there's still some progress. So I decided to bump to 2.01.0 % then. Just as a reminder for myself: it was the day when I watched Jacob Collier % perform \LUA\ (feat. MARO) live on YouTube (of course that is not about the % language at all, but still a nice coincidence). Just for the fun of it the number % bumped a few more times, just to catch up, so end 2019 we're at 2.03.5. % % Thanks to the patient \CONTEXT\ users we were able to apply the new macro scanner % and protection mechanisms that were introduced mid 2020.