Sun 09 Jul 2017 11:36:03 PM UTC, comment #7:
I think that was the last of the regressions of 3286.
|
Sun 09 Jul 2017 10:39:15 PM UTC, comment #6:
That last patch was pretty messed up -- hard to make much sense of it.
I did see that it touched on the two issues you mentioned.
I had to do a lot of manual fixups before, I just missed this last patch (or got fatigued).
Anyway: Please check out the formation of IJ and ij ligature in small caps,
and note that the OS/2 character ranges table is no longer automatic.
(By the way, the reason for that is that different FontForge versions populate this table differently, and I am forced to switch between different versions in avoidance of their peculiar bugs.)
The last and scariest trick to get the IJ thing going was to re-order the Latin lookups to be like they are in Sans Bold.
I hope this doesn't make anything in this face any worse than they are in that face.
Let me know if you see any further problems.
|
Sun 09 Jul 2017 06:57:54 AM UTC, comment #5:
Hi David,
Thanks, I had somehow missed your heads-up of May 19.
Fortunately, in the file I posted, I made pretty extensive notes of what I though I was doing with each patches. Indeed no patch after the 3242-3245 was applied, the reason being that something in that patch fouls up subsequent patches.
In the "procedure" notes, I indicated that I didn't thing 3287 was important, and that 3286 was the unsuccessful attempt to perform a regression. That explains what you're seeing.
But now I look into the patch file for that last commit, and besides the ij changes, I see other little tweeks. I should review that. As to why I thought it was unimportant, it ends with a lot of validations -- which are unimportant. Maybe I only saw those.
It isn't a huge patch but it contains several meaningful changes. Besides that I tweaked a width, adjusted the anchors on macron, and made several alterations to lookups.
This will take some time to sort out.
|
Sun 09 Jul 2017 01:53:57 AM UTC, comment #4:
This is just a reminder that there are still regressions from revision 3420. The longer the wait before fixing them, the harder they will be to fix.
|
Thu 18 May 2017 11:02:09 PM UTC, comment #3:
I’ll look at this in more detail later, but here’s what I’ve noticed. The most recent revision to FSaBO (3420) reverts all the changes since revision 3286; luckily there are only two: 3287 and 3340. It removes the OS/2 code page and Unicode ranges, which most of the other fonts include. It removes the substitutions from IJ and ij to sc.ij.
|
Thu 18 May 2017 08:46:14 PM UTC, comment #2:
OK this is like the fourth time I've been through this.
The results may not be perfect, but they're better than before, I think.
It looks as if the original problem way back at rev. 3105 looks like an inadvertent re-encoding to "Custom". That in itself is bad, if only that it knocks this face out of synch with the other faces. Anyway, other bad things happened in between. Then I tried to fix it, and made matters worse.
This version seems to have the letters you mentioned, and auxiliary letters where they were expected to be.
I balled up the files I used to build this (attached), with notes indicating what I did by hand, etc. in "procedures".
Thanks for pointing this out, David.
I should make a script to check for such nonsense -- but... how do I know what horseplay I'll get into next?
(file #40735)
|
Mon 31 Oct 2016 02:40:51 PM UTC, comment #1:
OK then I screwed it up.
The intent was to skip a single commit where something had gone wrong, but this sounds like several commits.
Thanks David.
|
Mon 31 Oct 2016 01:44:30 PM UTC, original submission:
Revision 3286 introduced some regressions to FreeSans Bold Oblique.
U+A7FA LATIN LETTER SMALL CAPITAL TURNED M should look like a Cyrillic small sha, but it now looks like a turned small capital M.
The following characters were removed.
- U+05BA HEBREW POINT HOLAM HASER FOR VAV
- U+05C5 HEBREW MARK LOWER DOT
- U+05C6 HEBREW PUNCTUATION NUN HAFUKHA
- U+05C7 HEBREW POINT QAMATS QATAN
- U+1D6B LATIN SMALL LETTER UE
- U+218A TURNED DIGIT TWO
- U+218B TURNED DIGIT THREE
- U+25CC DOTTED CIRCLE
- U+A7B0 LATIN CAPITAL LETTER TURNED K
- U+A7B1 LATIN CAPITAL LETTER TURNED T
- U+A7B6 LATIN CAPITAL LETTER OMEGA
- U+A7B7 LATIN SMALL LETTER OMEGA
- U+FE00 VARIATION SELECTOR-1
- U+FE01 VARIATION SELECTOR-2
- U+FE02 VARIATION SELECTOR-3
- U+FE03 VARIATION SELECTOR-4
- U+FE04 VARIATION SELECTOR-5
- U+FE05 VARIATION SELECTOR-6
- U+FE06 VARIATION SELECTOR-7
- U+FE07 VARIATION SELECTOR-8
- U+FE08 VARIATION SELECTOR-9
- U+FE09 VARIATION SELECTOR-10
- U+FE0A VARIATION SELECTOR-11
- U+FE0B VARIATION SELECTOR-12
- U+FE0C VARIATION SELECTOR-13
- U+FE0D VARIATION SELECTOR-14
- U+FE0E VARIATION SELECTOR-15
- U+FE0F VARIATION SELECTOR-16
|