![](https://secure.gravatar.com/avatar/46562c7d27388490d62e8d7a5bc960dd.jpg?s=120&d=mm&r=g)
Hi, I uploaded a new version with some new features discussed at the end of this mail. - A next version of luatex will break context (format making etc) but the change is unavoidable because of other macro package demands wrt security (context does that differently). The low level start up command has been adapted so unless you update indepdently thinsg should work ok. Of course this has no consequences for lmtx. - I've done a bit of callback cleanup (like getting rid of some mkiv luatex left-overs). It should have no effect on regular usage. - There again a few improvements wrt math (this time dealing with accents artifacts in some fonts). We check all the free math fonts for potential issues and then try to fix it (read: deal with it). It's a lot of work but we're getting there. - A few reported buglets have been fixed. - The biggest addition is however a new number model. In metapost one can now use 'posit' (positfun instance) as number mode which will use (currently 32 bit but at some point 64 bit) posit unums. It's also bit of a playground because 32 bit posits are not more accurate than 64 bit doubles. It anyhow exposed some potential issues that we'll look into. - There's also a (again currently 32 bit) posit library at the Lua end as playground. - However, 32 bit posits come in handy at the tex end because they pack better accuracy in 32 bits than floats and it makes it possible to have native floats, so we have: \newfloat (uses \floatdef, register like \dimendef is used in \newdimen) \newposit (uses \positdef, like \dimensiondef etc) \floatexpr (like \numexpr) \iffloat (like \ifnum) (also \ifabsfloat and i'll add the others too) Actually it adds an extra range of registers so it bumps memory usage a bit. I'm considering lowering the number of registers (currently 64K per category) because after all, if 64K is not enough then we can as well stick to 32K or 16K (for some even less). There's always Lua for unlimited storage of quantities. So, \newfloat\fooA \scratchdimen=123.456pt\relax [\type{dim:} \the\scratchdimen==123.456pt]\par \fooA =123.456\relax [\type{pos:} \the\fooA ==123.456] \par \fooA =\scratchdimen\relax [\type{pos:} \the\fooA ==123.456] \par \scratchdimen=\fooA\relax [\type{dim:} \the\scratchdimen==123.456pt]\par \scratchdimen\dimexpr\fooA+10pt\relax[\type{dim:} \the\scratchdimen==133.456pt]\par produces [dim: 123.456pt==123.456pt] [pos: 123.45600128173828==123.456] [pos: 123.45599365234375==123.456] [dim: 123.456pt==123.456pt] [dim: 133.456pt==133.456pt] As you can see dimens will get mapped onto floats not as integers (scaled points) but in pt units. Counts stay numbers. I need to handle glue too (not yet done). In the future I might use this for fractions but there might be some rough ends so for now it's experimental. Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | www.pragma-ade.nl | www.pragma-pod.nl -----------------------------------------------------------------