| <<<Back 1 day (to 2019/10/30) | Fwd 1 day (to 2019/11/01) >>> | 20191031 |
HenryS | kens I don't think any further investigation is needed. The selectable characters are as predicted. | 14:12.55 |
kens | OK | 14:13.06 |
| We might be able to do a better job, but I'd need to look into why we#'re getting a type 3 now | 14:13.30 |
HenryS | if you are curious, all I did was copy the NimbusMonoPS*.ttf over the NimbusMono fonts, you have to rebuild cleanly (romfs) | 14:18.25 |
| The PS variants have the necessary symbols. | 14:18.48 |
| a code change to the font table might have been easier, but... | 14:19.47 |
kens | Hmm | 14:40.23 |
| Well maybe when I have a spare minute (ho ho ;-) | 14:40.34 |
HenryS | also they surely won't be a customer. | 14:44.48 |
| it would be nice to have all this working though, all the information is there when internal fonts are in use. | 14:45.31 |
| but low priority | 14:45.42 |
kens | Maybe, I'd have to go poking to find out why we are even using a type 3 font, I would not have expected us to | 14:45.59 |
| Also, I'm not clear at all that we have the Unicode code point from a TrueType font | 14:46.18 |
| Unless we are using a £,1 CMAP subtable | 14:46.36 |
| 3,1 | 14:46.40 |
HenryS | All the characters I checked did have unicode values in the font, I'm pretty sure they all do. Yes the CMAP subtable is the issue | 14:47.11 |
chrisl | We can't reliably use a reverse cmap to go GID to Unicode | 14:47.56 |
HenryS | I thought something else was wrong too, but I'd have to look at it again. | 14:47.56 |
kens | Its been along time and I don;t remember off the top of my head, ut its worth noting that the TreuType embedded fotns have a ToUnicode CMap | 14:49.38 |
| But the type 3 does not | 14:49.44 |
HenryS | but we do have the entire TT font. | 14:49.51 |
kens | At the interpreter end yes | 14:50.07 |
| But pdfwrite isn't using it for the Greek characters | 14:50.15 |
| Its rendering to a bitmap and making a type 3 font | 14:50.22 |
HenryS | anyway like I said nice to have a solution to this but low priority | 14:51.16 |
kens | I wasn't planning to do anything till I've finished halftones and then looked at Seal's thing | 14:52.15 |
chrisl | Do the PCL fonts have 3.1 cmaps, and is that what we use? | 14:52.31 |
kens | does not remember | 14:52.42 |
HenryS | I thought last time I looked at this the only way to get pdfwrite to do this properly was to treat PCL fonts as CID fonts, but I might be mistaken | 14:53.46 |
kens | CIDFonts seems like overkill | 14:54.13 |
HenryS | kens how would you support all the characters in the font? Swap in different encodings? | 15:16.46 |
kens | Can PCL have more than 255 characters per fotn ? | 15:17.05 |
HenryS | yes, you can access a font by unicode in pcl | 15:17.40 |
kens | Then we'd need mulitple fonts. So that's probably why we're ending up in a type 3 | 15:18.04 |
HenryS | and all the symbol sets map to unicode | 15:18.08 |
kens | What kind of Unicode ? UTF-32 ? | 15:18.31 |
HenryS | 16 | 15:18.37 |
chrisl | Type 3 can still have a ToUnicode, can't they? | 15:18.54 |
kens | Well, we could at least use teh Unicode as the Unicode value | 15:19.00 |
| chrisl yes | 15:19.05 |
| Tha'ts what I was coming to :-) | 15:19.11 |
chrisl | Easy then :-) | 15:19.18 |
kens | We could use teh Unicode value as the ToUnicode, then we'd get teh copy/paste/search right | 15:19.29 |
| then it might look crap but at least that would work | 15:19.43 |
chrisl | The ideal solution would be to have all the glyphs in one fontfile, and instantiate it with different encodings | 15:19.56 |
kens | Yeah that's probably not going to fly with pdfwrite currently | 15:20.23 |
HenryS | and that's easier than using CID support in pdfwrite? | 15:20.29 |
kens | It assumes an output font (as opposed to CIDFont) will not have more than 255 glyphs | 15:20.39 |
| pdfwrite has CIDFont support | 15:20.54 |
| But PCL doesn't use CIDFonts | 15:20.59 |
HenryS | that's what I mean would be easier make PCL present the fonts as CID fonts to pdfwrite | 15:21.34 |
chrisl | Synthesizing a CIDFont could be.... interesting | 15:21.38 |
kens | Sounds hard to me | 15:21.48 |
chrisl | We'd have to synthesize a Unicode -> GID CMap | 15:24.32 |
| For each font | 15:24.56 |
kens | Hmm, well we know the Unicode and somewhere we must know the GID | 15:25.01 |
| Since we pull the glyph out of the font | 15:25.14 |
| s/we/PCL/ | 15:25.30 |
chrisl | The cmap table | 15:25.35 |
kens | It sounds plausible, but it also seems like a lot of work | 15:25.56 |
| Retrieving the Unicode code point for the ToUnicode CMap seems like an easier first step on the road | 15:26.22 |
chrisl | It would be - remember there's the crazy format that CMap data ends up in | 15:26.23 |
| <<<Back 1 day (to 2019/10/30) | Forward 1 day (to 2019/11/01)>>> | |