- dev-luatex - NTG Mailing Lists

nodelist based tex.saveboxresource
by Marcel Krüger 05 Aug '18

05 Aug '18

Hello, I want to suggest extending the tex.saveboxresource interface to accept a hlist or vlist node as an alternative to a box number as first argument. Currently creating a boxresource from Lua node list requires saving the list in a TeX box register, only to delete that register again by tex.saveboxresource. A direct way to transform a h/vlist into a boxresource would make such code easier and avoid potential side effects from changing the TeX registers. An experimental patch with a suggested implementation is attached. I also attached a patch implementing a node.scan method which I think would be another nice addition for LuaTeX: It scans a box with rules similar to `\setbox0` and returns the node. This allows passing box-like material to LuaTeX without requiring box registers. Best regards Marcel

4 4

socket.url addition
by Josef Kleber 19 Jul '18

19 Jul '18

Hi, in addition to Reinhard's report I want to report another change in luatex's behavior. In getmap.sty I now have to add \ifluatex\directlua{os.setlocale('','ctype')}\fi% in order not to run into strange encoding issues. See https://tug.org/pipermail/tex-live/2018-June/041997.html ff. Thread is ripped! :-( This was not neccessary with older versions of luatex, for sure not with v0.85, because I had to adapt getmap.sty for this version. Maybe you have an idea why this change is needed now. Best, Josef

2 1

socket.url not recognized as a built-in library
by Reinhard Kotucha 17 Jul '18

17 Jul '18

Dear Sirs, the script texlive/2018/texmf-dist/scripts/getmap/getmapdl.lua doesn't work anymore. No problems with former versions of TeX Live. It contains the lines local http = require("socket.http"); local ltn12 = require("ltn12") local url = require("socket.url") The luatex version in TL-2018 requires that I change the last line, and *only* the last line to local url = socket.url Otherwise luatex searches socket.url (and only socket.url) in Lua's search path (/usr/local/share/...) without success, though socket.url is already compiled into the binary. Because only socket.url is affected and not socket.http or ltn12 and everything worked in the past, I suppose that socket.url is not registered as a built-in package aymore in current releases of luatex. It seems that package.loaded.socket.url._NAME and package.loaded.socket.url._PACKAGE were removed accidentally. Regards, Reinhard $ diff -u socket.url-TeX-Live-2017 socket.url-TeX-Live-2018 --- socket.url-TeX-Live-2017 2018-07-17 02:57:49.894030637 +0200 +++ socket.url-TeX-Live-2018 2018-07-17 02:58:28.469719914 +0200 @@ -1,6 +1,4 @@ -package.loaded.socket.core.url._NAME = "socket.url" -package.loaded.socket.core.url._PACKAGE = "socket." -package.loaded.socket.core.url._VERSION = "URL 1.0.2" +package.loaded.socket.core.url._VERSION = "URL 1.0.3" package.loaded.socket.core.url.absolute() package.loaded.socket.core.url.build() package.loaded.socket.core.url.build_path() @@ -8,9 +6,7 @@ package.loaded.socket.core.url.parse() package.loaded.socket.core.url.parse_path() package.loaded.socket.core.url.unescape() -package.loaded.socket.url._NAME = "socket.url" -package.loaded.socket.url._PACKAGE = "socket." -package.loaded.socket.url._VERSION = "URL 1.0.2" +package.loaded.socket.url._VERSION = "URL 1.0.3" package.loaded.socket.url.absolute() package.loaded.socket.url.build() package.loaded.socket.url.build_path() -- ------------------------------------------------------------------ Reinhard Kotucha Phone: +49-511-3373112 Marschnerstr. 25 D-30167 Hannover mailto:reinhard.kotucha@web.de ------------------------------------------------------------------

2 1

ligature in \localleftbox \localrightbox
by Hironobu Yamashita 19 May '18

19 May '18

Hi, LuaTeX does not handle ligature inside \localleftbox / \localrightbox. (I found that \localleftbox{\hbox{<<~}} is OK, but ...) The following source reproduces the problem: \hsize=300pt \font\x=ec-lmr10\x {<<~\localleftbox{<<~}The original problem to be solved was for fine French typesetting, in which guillemets are placed running down the left side of a paragraph, as in this paragraph, so long as material is being quoted.~>>} Since \TeX\ breaks paragraphs in arbitrary places, it was impossible to develop a robust macro package that could, in a single pass, place the guillemets in the right positions. \bye The source compiles fine with Aleph. Is this change intended for LuaTeX? Best regards, Hironobu Yamashita

2 3

Proposal for new primitives
by Marcel Krüger 03 May '18

03 May '18

Hi, i want to propose two new primitives for LuaTeX: - The existing `\luafunction` has some similarities to `\count`, `\dimen`, `\skip`. Especially for most use cases, you will have to allocate a number, but there is no equivalent to `\countdef` etc. I think LuaTeX would benefit from `\luafunctiondef`, such that e.g. `\luafunctiondef\something1` would define `\something`, such that `\something` is equivalent to `\luafunction1`. What is the difference to creating a regular macro? This would allow `\something` to be expandable in a single step, which is especially important for emulating primitives from other engines with lua code. - Additionally it would be useful to allow "expandable temporary macro definitions". So for example \scantexmacro#1#2\stop{-#1-#2-}abc\stop def would expand to -a-bc-def This allows using the TeX macro argument parser inside expandable macros, especially when the delimiter can change. For example if you implement a property list as `\@key key1\@value value1\@key key2\@value value2...`, you can create an expandable macro to ge the value of key `#2` without requiring recursive calls with \def\propget#1#2{% \scantexmacro##1\@key#2\@value##2\@key##3\stop{% ##2% }#1\@key#2\@value\@key\stop } Another use-case would be implementingan expandable version of functionality similar to the expl3 `\seq_map_inline:Nn`. What do think? Is there any chance to see something like this in LuaTeX? I attached a first implementation as patch against the current experimental branch. Best regards Marcel Krüger

2 1

\primitive for non-primitive
by Hironobu Yamashita 28 Apr '18

28 Apr '18

Hi, LuaTeX raises "Missing primitive name" error for the following: \primitive\NONE \end PDFTeX (\pdfprimitive) and XeTeX (\primitive) simply ignores non-primitive, as the PDFTeX manual says > This command executes the primitive meaning of the following control sequence, > regardless of whether the control sequence has been redefined or made undefined. > If the primitive was expandable, \pdfprimitive expands also. > On the other hand, if the following control sequence never was a primitive, > nothing happens and no error is raised. > (In some versions of pdfTEX prior to 1.40.19, an error was wrongly given.) Hironobu Yamashita

2 3

Luatex 1.07.0 announcement
by luigi scarso 19 Apr '18

19 Apr '18

============================================================== LuaTeX 1.07 2018-01-17 ============================================================== (1) This release brings Lua 5.3 to LuaTeX. We have tested an experimental version for a few months and not encountered issues. (2) If you use Lua 5.3 you should be aware of a few changes in the language and its implementation. Numbers are now either integers or floats but are cast automatically. (3) In some cases (e.g. string.format) there is a more strict check. A float always has a decimal appended when serialized, so 1 becomes 1.0 as string. (4) We keep supporting the bit32 library on top of the new bit operators. Be aware of the fact that currently LuajitTeX does not have these operators. (5) Performance of LuaTeX with Lua 5.3 can be slightly better than with 5.2 but this really depends on your usage of Lua. In practice there should be no slowdown. (6) We are aware of possible substantial slowdown of string.format depending on the platform and compiler. This depends on Lua too and will be resolved in due time. (7) If you want to test with Lua 5.3 then have to build LuaTeX with --lua53. Because ConTeXt will switch to 5.3 after this release there will be binaries available at the ConTeXt garden. (8) The 2018 TeXlive distribution will ship with LuaTeX using Lua 5.2 but there will likely be an additional binary LuaTeX53 available. (9) From 2019 TeXlive will ship with LuaTeX and Lua 5.3 as default. We provide --lua52 as build flag for a while. By then we might be looking at Lua 5.4 already. (10) There are the usual patches and cleanups and a few small additions (mentioned in the manual) but there have been no fundamental changes in the API (apart from removing depricated and undocumented functionality). The LuaTeX team

1 0

openin_any = p in texmf.cnf
by Reinhard Kotucha 18 Mar '18

18 Mar '18

excerpt from texmf.cnf: ------------------------------------------------------------------------- % Do we allow TeX \input or \openin (openin_any), or \openout % (openout_any) on filenames starting with `.' (e.g., .rhosts) or % outside the current tree (e.g., /etc/passwd)? % a (any) : any file can be opened. % r (restricted) : disallow opening dot files % p (paranoid) : as `r' and disallow going to parent directories, and % restrict absolute paths to be under $TEXMFOUTPUT. openin_any = a ------------------------------------------------------------------------- Dear Sirs, when I set openin_any = p then fmtutil can't create any LuaTeX-based format files anymore: $ fmtutil --sys --all [...stuff omitted...] fmtutil [ERROR]: running `luatex -ini -jobname=luatex -progname=luatex luatex.ini </dev/null' return status 1 fmtutil [ERROR]: return error due to options --strict fmtutil [ERROR]: running `luajittex -ini -jobname=luajittex -progname=luajittex luatex.ini </dev/null' return status 1 fmtutil [ERROR]: return error due to options --strict fmtutil [ERROR]: running `luatex -ini -jobname=lualatex -progname=lualatex lualatex.ini </dev/null' return status 1 fmtutil [ERROR]: return error due to options --strict fmtutil [ERROR]: running `luatex -ini -jobname=dvilualatex -progname=dvilualatex dvilualatex.ini </dev/null' return status 1 fmtutil [ERROR]: return error due to options --strict fmtutil [ERROR]: running `luatex -ini -jobname=pdfcsplain -progname=pdfcsplain -etex csplain.ini </dev/null' return status 1 fmtutil [ERROR]: return error due to options --strict fmtutil [ERROR]: running `luatex -ini -jobname=dviluatex -progname=dviluatex dviluatex.ini </dev/null' return status 1 fmtutil [ERROR]: return error due to options --strict Only LuaTeX-based formats are concerned, all other format files are built successfully. Everything works as expected with openin_any = a and openin_any = r I'm using the latest stuff in tlpretest. Regards, Reinhard -- ------------------------------------------------------------------ Reinhard Kotucha Phone: +49-511-3373112 Marschnerstr. 25 D-30167 Hannover mailto:reinhard.kotucha@web.de ------------------------------------------------------------------

3 3

Semi-official git mirror of luatex repository
by Mojca Miklavec 12 Mar '18

12 Mar '18

Hi, I would be super grateful if there was a semi-official git mirror of the luatex repository that would also be updated via regular (hourly?) cron jobs. Henri would like to set up some Travis builders for testing LuaTeX builds as part of the effort to get better support for musl. There are some existing clones, some better than the others, some updated less frequently than the others. Does anyone volunteer to create a nice clone with regular updates, proper author map and some reasonable branch/tags setup? (Having it perfect is not really required, but it would be super nice to have a nice regularly updated clone.) Thank you very much, Mojca

2 2

Unknown command "mv -v" on Solaris
by Mojca Miklavec 07 Mar '18

07 Mar '18

Hi, LuaTeX build fails on Solaris due to "mv -v" in build.sh: if [ "$BUILDLUA52" = "FALSE" ] && [ "$BUILDLUA53" = "TRUE" ] then mv -v "$B"/texk/web2c/$LUATEXEXE53 "$B"/texk/web2c/$LUATEXEXE ls -l "$B"/texk/web2c/$LUATEXEXE fi mv: illegal option -- v Usage: mv [-f] [-i] f1 f2 mv [-f] [-i] f1 ... fn d1 mv [-f] [-i] d1 d2 build/texk/web2c/luatex: No such file or directory Can you please perhaps store the command and echo it before executing it instead of calling "mv -v" which is apparently not portable enough? Thank you, Mojca

2 1