Add build script

[davelab6]

https://github.com/TypeNetwork/fb-Decovar/blob/master/sources/2-build/Decovar.designspace
exists but doesn't appear referenced in any build script (https://github.com/TypeNetwork/fb-Decovar/search?utf8=%E2%9C%93&q=Decovar.designspace&type=)

(Whereas in Amstelvar it is in
https://github.com/TypeNetwork/fb-Amstelvar/blob/a879f3fbcc93826609e0d2bda6a90f0df0748cce/scripts/build.sh#L1)

[djrrb]

I’ve added one similar to Amstelvar. #24

Note that this does not cover the manual addition of a STAT table. Hopefully this will be added to fonttools varLib soon, so maybe not worth building in yet?

[davelab6]

I guess just add a comment to the build script to show where the manual step occurs.... although AFAIK its like... you TTX the binary, copy the stat table into the directory with the right filename, patch the index TTX file, and re-run ttx on it to zip the XML back into the binary... so that seems scriptable

[djrrb]

totally scriptable...do we need parity between amstelvar and decovar build processes? I would do this in python instead of a shell script...

[davelab6]

I don't need parity, I want a random type designer to be able to take the source files and reproduce the published binary with no fuss :) But that is rather ideal, and all that the contract requires is that you provide what you actually used to make your binary yourself, and adequate documentation for someone to reproduce what you did.

Section 15.2.2 provides more context if you want to nerd out on this stuff :) https://copyleft.org/guide/comprehensive-gpl-guidech16.html

[sberlow]

So a python script is ok?

[davelab6]

So a python script is ok?

Absolutely! waf and SCons are the popular Python build systems, or you call roll your own in "make file style" (as in sphinx-doc/sphinx#3196), or just have a simple imperative python script file...

[djrrb]

i'll make something basic tomorrow and we can always add a build system around it...not very knowledgeable about these different approaches

[davelab6]

Thank you!! (Simple is good :)

…

On Nov 14, 2017 7:25 PM, "David Jonathan Ross" ***@***.***> wrote: i'll make something basic tomorrow and we can always add a build system around it...not very knowledgeable about these different approaches — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#23 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAP9y7S8ndAcqZeIGGMVbVDc3hLH945mks5s2i9_gaJpZM4QcXw_> .

[djrrb]

Well, the good news is that the latest fontTools now writes its own STAT table, so we no longer need to include a more complex build script to address it. :-)

The bad news is that the build script is finding some incompatibilities and not generating the VF as expected. So just wanted to let you know I’m still on top of automating the build process, but it might take a bit longer than expected, and the changes will affect the sources and not only the script itself.

[davelab6]

Ok cool, I appreciate you looking into it

…

On Nov 15, 2017 8:41 AM, "David Jonathan Ross" ***@***.***> wrote: Well, the good news is that the latest fontTools now writes its own STAT table, so we no longer need to include a more complex build script to address it. :-) The bad news is that the build script is finding some incompatibilities and not generating the VF as expected. So just wanted to let you know I’m still on top of automating the build process, but it might take a bit longer than expected, and the changes will affect the sources and not only the script itself. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#23 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAP9y0wsjpzGDY5Gikc5yEmq26VpqsUMks5s2uoLgaJpZM4QcXw_> .

[dberlow]

Change the sources?

…

Sent from my iPad

On Nov 15, 2017, at 8:41 AM, David Jonathan Ross ***@***.***> wrote: Well, the good news is that the latest fontTools now writes its own STAT table, so we no longer need to include a more complex build script to address it. :-) The bad news is that the build script is finding some incompatibilities and not generating the VF as expected. So just wanted to let you know I’m still on top of automating the build process, but it might take a bit longer than expected, and the changes will affect the sources and not only the script itself. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

[djrrb]

no design changes...just to make them compatible or whatever needs doing to get Decovar to build from source without manual intervention

[dberlow]

Who said design changes?;) What I’m interested in are the source changes required for a style to properly build a stat table, if that is not already set up by nature and nurture in most fonts. Decovar will always be an alien.

…

Sent from my iPad

On Nov 15, 2017, at 9:22 AM, David Jonathan Ross ***@***.***> wrote: no design changes...just to make them compatible or whatever needs doing to get Decovar to build from source without manual intervention — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

[djrrb]

Haha ok good.

A basic STAT table is now being built by fontTools varLib (see commit here), so should now be included in all versions built with the latest fontTools.

There may be more source changes in store once more STAT-related data is specified in the designspace file (see discussion here)

[djrrb]

I think this is ready to roll, see #25. Please review, merge, and close if this looks good.