On 10/14/22 10:21, Hans Hagen via ntg-context wrote:

...

Hi,

I uploaded a new version. Apart from a few fixes / extensions already mentioned here (or known to those who wanted it) the main changes are in the math goodie files (work in progress, we will clean them up later) as part of improving the rendering of math.

Many thanks for the new version, Hans.

I have discovered that it includes in tex/texmf-context/source/luametatex what should be the LMTX source (about 11MB). It zips to 2 MB and totals to less of a picture on a fancy phone .. tre idea is to provide the user with all he needs as archive so no dependenccies (apart from a compiler). Also by including the source we can sort of guaranteed that you get what you expect to work with the tex files (no interference with distribution patches our of our control). It has always been part of the plan with luametatex. So .. it wil not be

On 10/14/2022 4:14 PM, Pablo Rodriguez via ntg-context wrote: optional. 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 -----------------------------------------------------------------

On 10/15/2022 10:48 AM, Pablo Rodriguez via ntg-context wrote:

On 10/15/22 09:41, Hans Hagen via ntg-context wrote:

...

On 10/14/2022 4:14 PM, Pablo Rodriguez via ntg-context wrote:

...

[...] Many thanks for the new version, Hans.

I have discovered that it includes in tex/texmf-context/source/luametatex what should be the LMTX source (about 11MB).

It zips to 2 MB and totals to less of a picture on a fancy phone ...

Many thanks for your reply, Hans.

At least on my system, I think every file is downloaded uncompressed.

With a low speed connection (here, the service isn’t great all the time), having to download over 400 files is a killer.

as has been pointed out, you only download the changed files and those are not many when sources are not included some complain, when they are others com-plain ...

The storage space on disk is irrelevant, I totally agree with you.

I wish I could say that large updates such as these weren’t a problem in some cases. But sometimes this isn’t true here.

With incremental updates, the downloading process takes way shorter in subsequent updates (as Alan pointed out). But the first time it might take too much (so the script quits).

just run twice as the script will pick up

...

the idea is to provide the user with all he needs as archive so no dependencies (apart from a compiler). Also by including the source we can sort of guaranteed that you get what you expect to work with the tex files (no interference with distribution patches our of our control). It has always been part of the plan with luametatex. So ... it will not be optional.

I asked to make this optional not to avoid having the source, but to be able to complete the update process.

i guess a few extra fonts are more demanding, actually the number of fonts in the installation dropped (for now) so that compensates the larger source tree

I’m not extremely confident, but I hope I will manage to update ConTeXt at the office (next working day). 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 -----------------------------------------------------------------

On 10/17/2022 7:12 PM, Pablo Rodriguez via ntg-context wrote:

On 10/15/22 11:28, Hans Hagen via ntg-context wrote:

...

[…] just run twice as the script will pick up

Hi Hans,

I have tried updating ConTeXt in Windows this morning and I’m afraid it hasn't worked.

Until I can manage this, I think I will download (to the extend that is possible [and if this works at all]) a Win64 distribution using Linux64 with "--platform=win64".

The problem is clearly not the speed of the internet connection. I use internet with the standard speed.

There are two things that strike me as possible causes.

1. Depending of which update is pending, the script announces a different number of files to be downloaded.

The problem is that each individual file download seems to require a complete connection.

I mean, it takes about ten seconds to establish a connection for each file. It is irrelevant how big the file actually is.

After that, it isn’t much harder to download the LMTX binary than a tiny text file.

so then just do a complete install

For most files, they aren’t downloaded in the first connection (they may require more than three downloads). It doesn’t seem to depend on size.

This is what makes it hard to reach even 10% in the download.

And even if some files may have been downloaded, once the script quits, the already downloaded files need to be downloaded again (since the main download process wasn’t completed).

2. The whole download is http instead of https.

sure, but if you have the curl(lib) installed in principle one could do a secure install (I didn't really test it but it's in there)

Sorry, but this may be causing the delay in the connections (because they may be being filtered by the company proxy [and I don’t want to ask]).

https://lmtx.pragma-ade.nl seems to be the only source used (in Windows) and its certificate is only valid for lmtx.pragma-ade.com (according to Firefox).

hm, you can change the url in the install script (maybe i can set up the server differently but not now)

Even using "./install.sh --secure --platform=win64" (I’m on Linux now, no Windows here), I get:

mtx-install | instance : install-lmtx mtx-install | platform : windows mtx-install | system : unix mtx-install | fetching 'http://lmtx.pragma-ade.com/install-lmtx//texmf.zip' […] mtx-install | fetching 'http://lmtx.pragma-ade.com/install-lmtx//texmf-context.zip'

I don’t know whether this is a bug or it is intended.

BTW, "local function checkurl()" on line 52 of mtx-install.lua ends with:

and find(s,"rotocols")

not important (p P)

I think it might read (although this doesn’t seem to change anything in my case):

and find(s,"protocols")

Is there anything that I could to to improve the download process? If per-file is slow at you end you can try to just install the whole lot (just delete the tex tree or at least the tma file that stores the hashes) because then the zip is downloaded and installed (one fetch).

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 -----------------------------------------------------------------

On 18/10/2022 09:40, Pablo Rodriguez via ntg-context wrote:

A complete install has revealed the real cause of the slow download: the antivirus software (that works like crap).

My antivirus software is called ... Unix ;-)

Hi, On Sun, Oct 16, 2022 at 6:07 PM Aditya Mahajan via ntg-context wrote:

On Fri, 14 Oct 2022, Hans Hagen via ntg-context wrote:

...

Hi,

I uploaded a new version. Apart from a few fixes / extensions already mentioned here (or known to those who wanted it) the main changes are in the math goodie files (work in progress, we will clean them up later) as part of improving the rendering of math.

Something is wrong with the placement of limits around an integral with NeoEuler:

\usetypescriptfile[euler]

\definetypeface[mainfont][rm][specserif][CharisSil][default] \definetypeface[mainfont][mm][math] [eulernova][default] \definetypeface[mainfont][tt][mono] [dejavu][default] [rscale=0.8, features=none] \setupbodyfont[mainfont,10pt]

\starttext \startTEXpage[offset=1mm] $\displaystyle \int_{0}^{1} f(x) dx$ \stopTEXpage \stoptext

gives the attached result.

This is because the integral "sits wrong" in its boundingbox. Almost all fonts have the glyph centered around the math axis, but there are a few that doesn't. In euler-math.lfg, add the tweak { tweak = "fixoldschool", }, In fact, we have not updated the euler goodie file for a while it seems. I think there are more things that can be improved. Hopefully before next release. It can also be mentioned that some fonts (Daniel Flipo was quick to fix concrete, erewhon and kpfonts) was fixed recently regarding this, and it is reported on and fixed in development of Lucida. /Mikael

On 10/17/2022 3:38 AM, Aditya Mahajan via ntg-context wrote:

On Sun, 16 Oct 2022, Mikael Sundqvist via ntg-context wrote:

...

Hi,

On Sun, Oct 16, 2022 at 6:07 PM Aditya Mahajan via ntg-context wrote:

...

On Fri, 14 Oct 2022, Hans Hagen via ntg-context wrote:

...

Hi,

I uploaded a new version. Apart from a few fixes / extensions already mentioned here (or known to those who wanted it) the main changes are in the math goodie files (work in progress, we will clean them up later) as part of improving the rendering of math.

Something is wrong with the placement of limits around an integral with NeoEuler:

\usetypescriptfile[euler]

\definetypeface[mainfont][rm][specserif][CharisSil][default] \definetypeface[mainfont][mm][math] [eulernova][default] \definetypeface[mainfont][tt][mono] [dejavu][default] [rscale=0.8, features=none] \setupbodyfont[mainfont,10pt]

\starttext \startTEXpage[offset=1mm] $\displaystyle \int_{0}^{1} f(x) dx$ \stopTEXpage \stoptext

gives the attached result.

This is because the integral "sits wrong" in its boundingbox. Almost all fonts have the glyph centered around the math axis, but there are a few that doesn't. In euler-math.lfg, add the tweak

{ tweak = "fixoldschool", },

This fixes the issue with eulernova, but not with pagellaovereuler. pagellaovereuler uses euler-with-pagella-math as a goodie file, but that is missing from the distribution. (I simply copied euler-math.lfg for my use case).

With tweak=fixoldschool, \startcases doesn't work correctly (see attached):

\usetypescriptfile[euler]

\definetypeface[mainfont][rm][specserif][CharisSil][default] \definetypeface[mainfont][mm][math] [eulernova][default] \definetypeface[mainfont][tt][mono] [dejavu][default] [rscale=0.8, features=none] \setupbodyfont[mainfont,10pt]

\starttext \startTEXpage[offset=1mm] \startformula \startcases \NC A \NC B \NR \NC C \NC D \NR \stopcases \stopformula \stopTEXpage \stoptext

Looks like some older experimental value is wrong: parameters = { -- DelimiterPercent = 901, DelimiterShortfall = 500, }, (that whole lfg is a todo) 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 -----------------------------------------------------------------

Hi Mikael, I use Lucida in my documents and did not notice any problem with the integrals and other large operators. Do you mean in the future I have to update to a new version of Lucida (and pay again…) or will future versions of LuaMetaTeX handle correctly large operators typeset in Lucida ? The issue I observed with the new upload of 2022.10.15 is that the presentations I typeset with the simpleslides module are broken after three pages, but could not set up a minimal working example to send to the list. Those presentations are typeset correctly with previous versions. Has Metapost changed some crucial settings ? Best regards: Otared

On 16 Oct 2022, at 19:03, Mikael Sundqvist via ntg-context wrote:

Hi,

On Sun, Oct 16, 2022 at 6:07 PM Aditya Mahajan via ntg-context wrote:

...

On Fri, 14 Oct 2022, Hans Hagen via ntg-context wrote:

...

Hi,

I uploaded a new version. Apart from a few fixes / extensions already mentioned here (or known to those who wanted it) the main changes are in the math goodie files (work in progress, we will clean them up later) as part of improving the rendering of math.

Something is wrong with the placement of limits around an integral with NeoEuler:

\usetypescriptfile[euler]

\definetypeface[mainfont][rm][specserif][CharisSil][default] \definetypeface[mainfont][mm][math] [eulernova][default] \definetypeface[mainfont][tt][mono] [dejavu][default] [rscale=0.8, features=none] \setupbodyfont[mainfont,10pt]

\starttext \startTEXpage[offset=1mm] $\displaystyle \int_{0}^{1} f(x) dx$ \stopTEXpage \stoptext

gives the attached result.

This is because the integral "sits wrong" in its boundingbox. Almost all fonts have the glyph centered around the math axis, but there are a few that doesn't. In euler-math.lfg, add the tweak

{ tweak = "fixoldschool", },

In fact, we have not updated the euler goodie file for a while it seems. I think there are more things that can be improved. Hopefully before next release.

It can also be mentioned that some fonts (Daniel Flipo was quick to fix concrete, erewhon and kpfonts) was fixed recently regarding this, and it is reported on and fixed in development of Lucida.

/Mikael ___________________________________________________________________________________ If your question is of interest to others as well, please add an entry to the Wiki!

maillist : ntg-context@ntg.nl / https://www.ntg.nl/mailman/listinfo/ntg-context webpage : https://www.pragma-ade.nl / http://context.aanhet.net archive : https://bitbucket.org/phg/context-mirror/commits/ wiki : https://contextgarden.net ___________________________________________________________________________________

Otared Kavian e-mail: otared@gmail.com Phone: +33 6 88 26 70 95

On 10/17/2022 8:43 AM, Otared Kavian via ntg-context wrote:

I use Lucida in my documents and did not notice any problem with the integrals and other large operators. Do you mean in the future I have to update to a new version of Lucida (and pay again…) or will future versions of LuaMetaTeX handle correctly large operators typeset in Lucida ?

I think there is some free upgrade policy. In general the old font should work ok given that one can live with the (current) inaccuracies (ance it's done there will be an article about the improvemnets that clarifies things). We already concluded long ago that the suite of math fonts is somewhat inconsistent (between and within fonts). We try to deal with that as good as possible and believe that we have found the right mixed approach. If needed we can catch averythign in the goodie files but we try to minimize it so once this is all settled down, we can improve specific font related issues based on {\em realistic} user mwe's as we go. Concerning large operators .. it worked before so ... the main reason why in e.g. euler it failed was a configuration option (euler fonts will be fixed too) which is a side effect of more advanced script anchoring in lmtx.

The issue I observed with the new upload of 2022.10.15 is that the presentations I typeset with the simpleslides module are broken after three pages, but could not set up a minimal working example to send to the list. Those presentations are typeset correctly with previous versions. Has Metapost changed some crucial settings ? some mp->tex interfacing macros were upgraded but that should not really matter much so i need an example of what goes wrong (quite some month ago we fixed some left/right page related interfacing but that should not affect single sided documents)

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 -----------------------------------------------------------------