On 18.07.21 22:42, Jordan Brown wrote:
Ah... I hadn't realized that the PR automated builds produced
persistent results that are publicly available.
Persistent is relative :-). IIRC they store the artifacts for
30 days. If needed we can re-trigger the build for another 30
days or stash them on the file server. But the idea is to merge
soon anyway, so maybe we don't need that.
ciao,
Torsten.
Ahh, I see. Thank you for explaining that.
On 7/18/2021 4:29 PM, Jordan Brown wrote:
On 7/18/2021 9:37 AM, Douglas Miller wrote:
I must be missing something here. When I try this code below, I get
the message
WARNING: Ignoring unknown function 'textmetrics', in file ./, line 2.
You are missing that textmetrics() is a not-yet integrated feature
that I'm working on; this is a preview of functionality that will soon
(fingers crossed) be available.
Sorry, I'm coming in late to this conversation (been busy this week).
So, is textmetrics() a remix of 2018 Alexander Pruss's fontmetrics?
https://www.thingiverse.com/thing:3004457
On Sun, Jul 18, 2021 at 3:47 PM Douglas Miller doug@milmac.com wrote:
Ahh, I see. Thank you for explaining that.
On 7/18/2021 4:29 PM, Jordan Brown wrote:
On 7/18/2021 9:37 AM, Douglas Miller wrote:
I must be missing something here. When I try this code below, I get the
message
WARNING: Ignoring unknown function 'textmetrics', in file ./, line 2.
You are missing that textmetrics() is a not-yet integrated feature that
I'm working on; this is a preview of functionality that will soon (fingers
crossed) be available.
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org
On 7/18/2021 2:08 PM, Leonard Martin Struttmann wrote:
So, is textmetrics() a remix of 2018 Alexander Pruss's fontmetrics?
https://www.thingiverse.com/thing:3004457
https://www.thingiverse.com/thing:3004457
Well, sort of. They have some of the same goals: give the program
information about how text would be laid out.
That program has, contained in it as OpenSCAD data, metrics for some 46
particular fonts. It looks like it probably has data for every
character in those fonts, but that's hard to check. It duplicates data
that is available as part of the font and is available to the OpenSCAD
internals. By its nature, it can only offer information about fonts
that it has specific support for.
This feature does not duplicate font metric data. Rather, it gives you
access to the font metric data that OpenSCAD already has internally. It
does not have data for some particular set of fonts; instead, since it's
getting the information from the font itself, it has data for every font
installed, now and in the future.
Does that make the picture clearer?
On 18.07.21 23:24, Jordan Brown wrote:
It does not have data for some particular set of fonts; instead,
since it's getting the information from the font itself, it has
data for every font installed, now and in the future.
Also in addition textmetrics() works on the actually shaped
glyphs, including ligatures for character combinations like "fi"
which is almost impossible to capture otherwise. While for some
languages like English, that's not too problematic, it's quite
critical for others e.g., Farsi (written in Persian calligraphy).
ciao,
Torsten.
On Sun, 2021-07-18 at 22:38 +0200, Torsten Paul wrote:
In the meantime it's possible to try the feature via the automated
builds for Windows and Linux.
Linux AppImage (x86/64-bit)
https://8305-1049088-gh.circle-artifacts.com/0/64-bit/OpenSCAD-2021.07.08.ai8305-_PR36842021.07.08.ai8305-x86_64.AppImage
Windows (32-bit)
https://8306-1049088-gh.circle-artifacts.com/0/32-bit/OpenSCAD-2021.07.08.ci8306-x86-32_PR3684.zip
https://8306-1049088-gh.circle-artifacts.com/0/32-bit/OpenSCAD-2021.07.08.ci8306-x86-32_PR3684-Installer.exe
Windows (64-bit)
https://8304-1049088-gh.circle-artifacts.com/0/64-bit/OpenSCAD-2021.07.08.ci8304-x86-64_PR3684.zip
https://8304-1049088-gh.circle-artifacts.com/0/64-bit/OpenSCAD-2021.07.08.ci8304-x86-64_PR3684-Installer.exe
I gather these will change over time. Is there a URL where we can find
the latest builds?
On 18.07.21 23:57, larry wrote:
I gather these will change over time. Is there a URL where we can find
the latest builds?
No, PR builds (things not integrated into the main application) don't
have a stable URL. Only the latest dev builds are automatically
downloaded and linked on the dev-snapshots page.
It's possible to get to the page via github by finding the last commit
on the PR, e.g. https://github.com/openscad/openscad/pull/3684
Currently that is this line:
finish textmetrics tests rename <green checkmark> 47f16b8
Clicking on the green checkmark opens a popup with actions, the
interesting ones are the ci/circleci ones, e.g.
ci/circleci: openscad-mxe-64bit <Details>
The details links jumps to CircleCI which unfortunately now requires
a login of some sort, any existing github login should do though.
The "Artifacts" tab should then show the download links if those
are still available.
ciao,
Torsten.
On 7/18/2021 2:47 PM, Torsten Paul wrote:
Also in addition textmetrics() works on the actually shaped glyphs,
including ligatures for character combinations like "fi" which is
almost impossible to capture otherwise. While for some languages like
English, that's not too problematic, it's quite critical for others
e.g., Farsi (written in Persian calligraphy).
Good point.
I suspect that Pruss's work will get the wrong results for "difficult"
in fonts with ligatures - Palatino, for instance - because it doesn't
know about the "ffi" ligature.
(BTW, before or after my work, the "spacing" parameter doesn't work
right with those cases. "spacing" doesn't work well with any
proportionally-spaced fonts.)
Here's "HarfBuzz", the font layout engine's name, as written in Persian,
with a textmetric-measured box around it:
Thanks for the detailed reply. I look forward to playing with textmetrics()
in the future.
On Sun, Jul 18, 2021 at 6:17 PM Jordan Brown openscad@jordan.maileater.net
wrote:
On 7/18/2021 2:47 PM, Torsten Paul wrote:
Also in addition textmetrics() works on the actually shaped glyphs,
including ligatures for character combinations like "fi" which is almost
impossible to capture otherwise. While for some languages like English,
that's not too problematic, it's quite critical for others e.g., Farsi
(written in Persian calligraphy).
Good point.
I suspect that Pruss's work will get the wrong results for "difficult" in
fonts with ligatures - Palatino, for instance - because it doesn't know
about the "ffi" ligature.
(BTW, before or after my work, the "spacing" parameter doesn't work right
with those cases. "spacing" doesn't work well with any
proportionally-spaced fonts.)
Here's "HarfBuzz", the font layout engine's name, as written in Persian,
with a textmetric-measured box around it:
OpenSCAD mailing list
To unsubscribe send an email to discuss-leave@lists.openscad.org