Apache OpenOffice (AOO) Bugzilla – Full Text Issue Listing
|Summary:||some default numbering bullets *not* in OpenSymbol|
|Status:||CLOSED FIXED||QA Contact:||issues@sw <issues>|
|Priority:||P3||CC:||frank.loehmann, frank.meies, hdu, issues, jr, lohmaier, michael.ruess, mmeeks, nospam, oo, rb.henschel, rodo, stefan.baltzer, tuharsky|
|Issue Type:||DEFECT||Latest Confirmation in:||---|
|Issue Depends on:||54450, 45163, 56252|
Description caolanm 2006-01-03 16:41:50 UTC
Referencing issue 42357, issue 59212 and issue 32327 and the spec at http://specs.openoffice.org/writer/numbering/NumberingBullet1.sxw shows that the default bullets for bulleted numbering levels 2,5 and 8 for writer are now all 0x25CB for 2.0.1. My problem is that as attachment one shows, there is no 0x25CB in OpenSymbol, leading to attachment two :-(
Comment 1 caolanm 2006-01-03 16:42:32 UTC
Created attachment 32872 [details] bullets in relevent area of opensymbol font
Comment 3 caolanm 2006-01-03 16:46:55 UTC
*** Issue 59212 has been marked as a duplicate of this issue. ***
Comment 4 caolanm 2006-01-03 16:48:32 UTC
cmc->mmeeks: fyi, for the love I know you have for OpenSymbol
Comment 5 caolanm 2006-01-04 09:49:40 UTC
Created attachment 32895 [details] how about this bullet instead
Comment 6 lohmaier 2006-01-04 15:37:14 UTC
what about adding that symbol to opensymbol instead? U+e572 is a bad choice since it is from the private use area of the font. This breaks compatibility (think of exporting to HTML, exporting to doc, etc. The target-machine won't have opensymbol installed, but there is a bigger chance that the user has a font with the bullets from the regular area. (Arial, Times New Roman, Courier New, ... all have the U+25cb bullet. Apart from that: U+e572 is not part of OpenSymbol either (none of the fonts I have installed has a bullet there) - so this would break vanilla OOo instead of fixing it.
Comment 7 michael.ruess 2006-01-05 08:22:10 UTC
MRU->FL: should the OpenSymbol font be enhanced or should we take different default characters for 2nd, 5th and 8th bullet level?
Comment 8 mmeeks 2006-01-05 10:45:13 UTC
So - Radek has been working on this and has extended our custom symbol font - based on re-arranging the existing OpenSymbol glyphs. It's at the normal place Caolan: http://go-oo.org/ooo-build/fonts/ Of course - Sun has traditionally fiercly resisted improving opens___.ttf, and without the trivial fixes / re-arrangements we provide in our version (available under the JCA etc.) all manner of other bullets are broken out of the box. It would be *really* nice if Sun would re-consider incorporating our changes: 2006-01-04 Radek Doulik <email@example.com> * fonts/OpenSymbol.sfd: added circle glyph to fix https://bugzilla.novell.com/show_bug.cgi?id=141377. generated new opens___.ttf using fontforge tool.
Comment 9 nospam 2006-01-05 11:22:46 UTC
IH: I think we should really consider using Radek's version of the font if it is better than the currently used one. I also think that this should/could be fixed for 2.0.2 and not in 3.0. It is important that this font works in OOo and we have/had so much trouble with this in the past and now and I would be very glad to have a working version of the font and I appreciate any kind of help with those issues here. Radek, what is about the other existing bugs we have in OpenSymbol (e.g. 54450 and 54182 or the incompatibility with some Win Systems), do you have fixes for this in your version? Do you have an idea how to add proper hinting information to OpenSymbol (unfortunately FontForge does not support this feature of adding truetype hinting to fonts)? Thanks...
Comment 10 firstname.lastname@example.org 2006-01-05 13:16:51 UTC
HDU->QA reminder: Please also test the deadly combination of (Win98) + (symbol font) + (unicode symbols) extensively. Also see issue 45163 (where we had to patch FontForge to get Win98 happy).
Comment 11 mmeeks 2006-01-05 14:22:37 UTC
> IH: I think we should really consider using Radek's version of the font > if it is better than the currently used one. That's great news - it sucks to have to have this in ooo-build; I'd love to remove it from our build. > It is important that this font works in OOo and we have/had so much > trouble with this in the past and now and I would be very glad to have > a working version of the font and I appreciate any kind of help with > those issues here. Nice. > Radek, what is about the other existing bugs we have in OpenSymbol > (e.g. 54450 and 54182 or the incompatibility with some Win Systems), > do you have fixes for this in your version ? Well - what ours fixes [ and I did much of the historic work here ] is big chunks of wingdings, webdings etc. that are missing equivalents in the positions specified in vcl's font mapping tables. ie. before we added glyphs at these locations we simply got '' (square box missing-glyph symbols) :-) Some good examples would be the 2 bullet tests at: http://go-oo.org/ooo-build/test/draw/ of course - a better way is to use =char(A1) in a spreadsheet & map out the columns for the various fonts. I did that in the past but can't find the test sheet for some reason. Wrt. breaking Win32 functionality - I guess that's unlikely AFAICS there is no code that checks whether a glyph is available in opens___.ttf and if not falls back to something more useful - but I may be mistaken. And of course Win32 machines all have the fonts - so the effects are seen mostly only on Linux machines. HTH.
Comment 12 nospam 2006-01-05 14:58:13 UTC
IH: what about copyrigth issues if there are characters/symbols of wingdings and webdings in the font? Does anyone here have a proper knowledge what is allowed/not allowed in this specific field?
Comment 13 email@example.com 2006-01-05 15:23:49 UTC
> Wrt. breaking Win32 functionality - I guess that's unlikely AFAICS As a counter example I'd like to point to the high priority issue 45163 and issue 40717. Also it is not a good idea to rely on fallback mechanisms. So the point I'm trying to make is: Test the new OpenSymbol font so extensively that it doesn't cause regressions.
Comment 14 firstname.lastname@example.org 2006-01-05 15:24:06 UTC
> Wrt. breaking Win32 functionality - I guess that's unlikely AFAICS As a counter example I'd like to point to the high priority issue 45163 and issue 40717. Also it is not a good idea to rely on fallback mechanisms. So the point I'm trying to make is: Test and fix the new OpenSymbol font so extensively that it doesn't cause regressions.
Comment 15 radekdoulik 2006-01-05 15:35:34 UTC
Unfortunatelly I don't have win98/MS Office here. I will attach the modified font so that someone with win98/MS Office can test it. Changes I made to out opens___.ttf: * added circle glyph * synced characters [names] with upstream opens___.ttf so that it has all the character the upstream version have. I will also attach output of sfddiff of these too. I used this version of fontforge (I am not sure whether it already contains your patch): rodo@aquarius:~/cvs/ooo-build/build/src680-m145> fontforge --version Copyright (c) 2000-2005 by George Williams. Executable based on sources from 13:45 3-Aug-2005. fontforge 20050803
Comment 16 radekdoulik 2006-01-05 15:39:12 UTC
Created attachment 32928 [details] output of sfddiff updated_font upstream_font
Comment 18 mmeeks 2006-01-05 16:09:54 UTC
> IH: what about copyrigth issues if there are characters/symbols of > wingdings and webdings in the font? There are not :-) clearly copying glyphs from propriatory fonts would be gross infringement. We havn't done that - instead, the primary task is just cut/paste/resize/flip etc. of the existing opens___.ttf glyphs to the correct positions by consulting the vcl font mapping tables. In addition - we created a small number of new glyphs - (some really simple ones - > >> that sort of thing - that required little effort). All our work in the font is covered by the JCA - so, AFAICS there should be no issue.
Comment 19 frank.loehmann 2006-01-05 17:13:09 UTC
FL: I made some investigations why we have not seen this OOo issue before integrating this change for 2.0.1. The spec. (see section 2.0.2): http://specs.openoffice.org/writer/numbering/NumberingBullet1_PP1.sxw shows the symbols formatted in OpenSymbol, but is seems that Writer did make an internal font substitution due to the missing 0x25CB symbol in OpenSymbol. I will double check this with help from our Writer development tomorrow. Personally I have no problem to use the changed OpenSymbol font even for StarOffice, because the 0x25CB symbol in the StarSymbol font looks pretty ugly due to its bigger size. I will discuss this with our font expert tomorrow. Furthermore OpenSymbol could be extended and adjusted and we are not allowed to change StarSymbol on our own due to it's license. I propose to fix this issue for 2.0.2.
Comment 20 email@example.com 2006-01-06 08:52:17 UTC
I would like to point out that as of today the CVS version of FontForge no longer triggers the OpenSymbol on Windows problems mentioned above. The idea from the patch in issue 45163 was commited there. So anyone doing artwork in OpenSymbol is advised to use at least a CVS-2006-01-06 version.
Comment 21 radekdoulik 2006-01-06 10:13:51 UTC
Created attachment 32953 [details] new updated font generated with fonforge build from today cvs sources
Comment 22 stefan.baltzer 2006-01-06 10:50:02 UTC
*** Issue 9437 has been marked as a duplicate of this issue. ***
Comment 23 nospam 2006-01-06 13:05:31 UTC
IH: I have installed Radek's latest version of the font on WIN XP and MAC OS X. It works fine in OOo on both platforms, but Windows System Utilitiy character map tool does not like the font - it just shows empty fields, this kind of behaviour was in the past also a good hint for problems with the font on Win98. Radek's font is based on our 1.7 Version - so we also need to try to incorporate our later fixes into this version. I have also seen that some symbols occur more than once in the font. Radek, is this becauso of mapping issues or is it just a mistake? Thanks...
Comment 24 frank.loehmann 2006-01-06 13:25:04 UTC
FL: I have just tested the OpenSymbol font on my Windows system, and I had problems in accessing it in the Insert-Special Characters dialog of my StarOffice 8. StarOffice 7 and the system viewer from Windows worked fine. Any ideas? I think the bullets, especially for level 1 and 2, should be of the same size and same positioning. The current bullets do not achieve these points. The StarSymbol font has the same issue and I think this does not look good.
Comment 25 nospam 2006-01-06 13:31:36 UTC
IH->FL: I can not reproduce FL's problem on my WinXP system. Insert->Special Character of OpenSymbol works fine in SO8 and OOo2.0. It looks like a Font-ID problem. FL, did you remove the StarSymbol font and what other Symbol fonts do you have on your system. What is about OpenSymbol in other Apps on your System - are there also problems?
Comment 26 frank.loehmann 2006-01-06 13:38:52 UTC
FL: Problem in SO8 solved after restart incl. quick starter. SO7 and other apps did not need a restart.
Comment 27 mmeeks 2006-01-06 14:39:46 UTC
>IH: I have also seen that some symbols occur more than once in the font. > Radek, is this becauso of mapping issues or is it just a mistake? Thanks... That is correct - this is deliberate, since similiar looking (but non-identical) glyphs occur across lots of Win32 fonts - eg. a round '*' bullet glyph; in opens___.ttf we tend to just copy the glyph around to the places where it should be - and not change it, since the differences are small for these glyphs between fonts.
Comment 28 radekdoulik 2006-01-06 15:37:04 UTC
> I think the bullets, especially for level 1 and 2, should be of the same size > and same positioning. The current bullets do not achieve these points. The > StarSymbol font has the same issue and I think this does not look good. I am playing with it now. I have made level 1 and 2 equal size. I have also changed version to 1.10 and updated copyright to 2006. BTW, as fontforge doesn't produce good hints intruction for ttf I am thinking about using PS type 1 font. How does this sound? So far I was unable to make OOo use type1 font, but other applications were able to use it and it looks way better for small sizes (as default bullet size here for example) Any idea why my OOo doesn't see/use the type1 font?
Comment 29 firstname.lastname@example.org 2006-01-06 16:22:45 UTC
> I am thinking about using PS type 1 font. How does this sound? Veto! Here are some reasons: Type1 fonts are not native on some platforms for OOo like Win95/98/ME/NT. Sometimes additional software must be installed. Truetype is more unicode friendly and easier to subset. Type1 is limited to 255 characters. ...
Comment 30 radekdoulik 2006-01-06 16:38:55 UTC
> Type1 fonts are not native on some platforms for OOo like Win95/98/ME/NT. > Sometimes additional software must be installed. With fontforge we can generate both formats and package type1 for unix distros and ttf for win systems. (with single .pfb source) > Truetype is more unicode friendly and easier to subset. > Type1 is limited to 255 characters. I didn't look into specs, but the font I generated and used here contains 1000+ characters and works OK in other apps with better looking output.
Comment 31 lohmaier 2006-01-06 18:42:59 UTC
apart from the "bullets should have same position and size", the new font unfortunately still increases the line-spacing way too much. Please apply the settings I mentioned in issue 54450 to fix this problem. It would be unfortunate if that one would not be fixed when a major QA regarding OpenSymbol is sheduled/needed anyway... * Choose Elements|Font-Information Tab OS/2 Tab Metrics * In the box "Win Ascent Offset" enter "-630" (old value "0") * In the box "Wind Descent Offset" enter "-200" (old value "0") leave the other settings unchanged.
Comment 32 philipp.lohmann 2006-01-09 10:35:16 UTC
radekdoulik: a type1 font can contain more than 256 glyphs, but the contained encoding vector can only contain 256 entries. This problem is usually circumnavigated by using different encoding vectors (e.g. you have multiple lines in the Xserver's fonts.dir files for the same font), but this method somewhat fails for symbol fonts.
Comment 33 nospam 2006-02-01 09:22:38 UTC
IH: it is getting so quiet here. Could the QA and UE people please decide how we will / have to proceed here? As I already mentioned my suggestion is to take Radek's font and add our changes from 1.7->1.9 into this one.
Comment 34 lohmaier 2006-02-02 21:47:03 UTC
if "Radek's font" is the attached one, please modify it to fix the linespacing problem as directed earlier before integrating it.
Comment 35 pchunter 2006-02-24 19:30:46 UTC
I do not know how your opensymbol font dependencies work for bullet display, but wouldn't it be easier to add a "change bullet" button in the gui that becomes active after a character is selected for format-bullets and numbering-bullets(tab) that would allow the user to define the character and size of the bullet (or make it a right click option for the default character or all eight and or all levels) that would seem to solve the issues in about a dozen posts I have read and eliminate the need to force opensymbol to be used here. then the user defines the character and size and they can do with it as they please. This is my one pet peave with open office. I can not define my default bullet and have the program store the character info. It seems the programming approach is to use a single font instead of making the character user defined.
Comment 36 andreas.martens 2006-02-27 15:45:06 UTC
You could create a template with your preferred numbering/bullets as soon as issue 56252 is fixed.
Comment 37 radekdoulik 2006-03-09 10:53:47 UTC
Created attachment 34703 [details] latest font with unified size of level 1 and 2 item bullets
Comment 38 radekdoulik 2006-03-09 10:58:49 UTC
Created attachment 34707 [details] source file to previous .ttf font
Comment 39 radekdoulik 2006-03-09 11:05:02 UTC
Created attachment 34708 [details] one more font with linespacing fixed
Comment 40 radekdoulik 2006-03-09 11:05:57 UTC
Created attachment 34709 [details] source file to previous .ttf font
Comment 41 radekdoulik 2006-03-09 11:07:18 UTC
I have attached new font with linespacing and level 1,2 bullet sizes fixed. Let me know if anything else is needed.
Comment 42 caolanm 2006-03-09 14:24:28 UTC
cmc->fl: So can we standardize on this font as the opensymbol font, and retarget for 2.0.3 ?
Comment 43 radekdoulik 2006-03-13 10:03:52 UTC
I fixed one more problem today, attaching new fonts. I changed the encoding to match the original upstream version of opens___.ttf and fixed problem with minussign glyph which had set unicode value to 0x00ad instead of 0x2212.
Comment 46 nospam 2006-03-13 15:06:56 UTC
IH: I have tested Radek's latest Version on WinXP and Mac OS X. The font works very good and even in Windows character map tool it looks ok (that is a good hint for compatibility with Win98, but this has to be tested explicitely, unfortunately I haven't got Win98). Good work, Radek. Thanks a lot.
Comment 47 nospam 2006-03-22 08:31:00 UTC
IH->Radek: We had some issues with OpenSymbol in the past because we could not include TrueType Hinting into the font. The issues were about some symbols (e.g. - or =) disappearing at specific zoom rates or point sizes. I fixed this in the other branch of the font by changing the font outline itself (I made some of the thin lines thicker). Do you have an idea how we will try to avoid such problems in your version of the font? Do we or you have any chance to put TrueType Hinting into the font or do you have any other idea or suggestion to fix this? IH->ALL: We still need QA and UserExp. approvement for setting this task to 2.0.3. Any news?
Comment 48 radekdoulik 2006-03-22 14:41:53 UTC
fontforge doesn't provide good truetype font hinting. I am not sure at what time Michael branched the font, but it might contain the changes you did I guess. If you remember the problematic glyphs, it is possible to find out which glyphs differ from original.
Comment 49 radekdoulik 2006-03-22 14:47:16 UTC
Created attachment 35133 [details] sfddiff output for original and new font
Comment 50 nospam 2006-03-22 15:08:41 UTC
IH->Radek: I never had any chances to save the truetype hinting instructions in a font in fontforge, neither good or bad ;-) I think the important chars we should look at are: - (minus) + (plus) and = (equal) I will also have a look at your diff file.
Comment 51 nospam 2006-03-22 15:14:46 UTC
IH: I just have tested the latest font in OOo's Formula Editor: plus, minus and equal chars look good but I have seen a problem with some brackets and some other symbols; their upper part is some kind of cut away. Can someone else confirm this?
Comment 52 nospam 2006-03-22 15:20:26 UTC
Created attachment 35137 [details] Screenshot of my formula editor problem
Comment 53 frank.loehmann 2006-03-22 16:17:51 UTC
FL: Here are my results after I did some investigation with help of the Writer team. Original problem of current issue: Character 0x25CB not found in OpenSymbol. This would be fixed by this issue. Maybe some regressions in Math or with other symbols due to font problems with the changed OpenSymbol font. Related problems not addressed today : 1. Sizes of bullets are to big (especially #2, #5, #8) 2. Sizes of bullets are not relative to font used in paragraph due to assigned â€œBulletâ€ character style) 3. Bullets (especially #2, #5, #8) do not look very good (fixed for OOo with this current issue, but still valid for SO due to use of different font) Other aspects: - old document must not change when loaded in newer version of OOo Discussed solutions: 1. Change sized of the characters in the OpenSymbol / StarSymbol font to get rid of the â€œBulletâ€ character style -> has been rejected 2. Change size attribute in â€œBulletâ€ character style from 9 to 8 or 7 pt -> does not solve problem #2 and #3. 3. Change size using hard coded size calculation -> high effort, maybe probs with old documents but would solve #1-2 4. Use new set of symbols -> would fix #1-3 if there are better ones in StarSymbol and OpenSymbol. But why we hadn't used those from the very beginning? Root cause analysis: We have defined wrong bullet characters in specification. Input on these characters used came from development. Nobody ever challenged the symbols used in the spec. Then change did not made it into OOo 2.0 final, because change got lost on integration in master (two times). Then integrated for PP1 too shortly before release of OOo 2.0.1, so problem had been overlooked. Furthermore StarSymbol font is present on all systems used @Sun, so OpenSymbol font didn't get used on in-house testing, even if OOo version was tested. Solution The solution of entire problem is very easy and i63395 handles this needed change. We need to use exactly the same bullets used after importing a Microsoft Word document. The font substitutionof OOo will handle systems where those fonts are not present. Then the round trip experience is good and even problem #3 will get fixed. Unfortunately this means that the changed OpenSymbol font of this issue will only be used when loading old OOo 2.0.0 - 2.0.1 documents. But nevertheless we need the current fix for OOo 2.0.3. to fix the missing character in those cases.
Comment 54 nospam 2006-03-22 16:57:52 UTC
IH->FL: Why is the changed font only used for 2.0 and 2.0.1 docs? Which font is used for the other ones in OOo? It can only by OpenSymbol but why not also the changed version? I do not understand this. Which fix do you mean by we need this fix? Nethertheless I think we should switch to Radek's new version of the font because it is better than all other OpenSymbol Versions - regardless what this means for our bullet problem. It does not look like that we can fix this whole bullet problem thing before 3.0, because for this we need a completely new font for SO and OOo with TrueType Hinting and so on...
Comment 55 frank.loehmann 2006-03-22 17:24:58 UTC
FL: OOo will use different characters as soon if 63395 get fixed. Then we will use Courier New, Symbol and Windings. If not present on the system, our font substitution determines a replacement (this is already working, see attached WW doc on Linux). If we will discover further issues, we still could use their equivalents from StarSymbol and OpenSymbol and could shift the fix to OOo 2.0.4. Radek's font will be integrated to OOo 2.0.3 that's why I have changed the target to OOo 2.0.3.
Comment 56 frank.loehmann 2006-03-22 17:27:51 UTC
Created attachment 35139 [details] Word standard bullet demo document
Comment 57 radekdoulik 2006-03-23 11:47:37 UTC
Radek -> IH: I am unable to reproduce the formula problem here. Please could you attach a document with the formula?
Comment 58 nospam 2006-03-23 15:05:21 UTC
Created attachment 35178 [details] File for the Formula Problem
Comment 59 nospam 2006-03-23 15:09:57 UTC
IH->Radek: I have attached the problematic formula, but I think the problem will always occur when brackets or something like that are used. My platform is WinXP without font smoothing and the zoom rate was set to 100%. IH->FL: If we use Courier New, Symbol or Windings as replacement fonts, what will happen on Linux?
Comment 60 frank.loehmann 2006-03-27 12:29:22 UTC
FL: Unfortunately my submission from Friday did not made it into this issue. So I give it a second try. FL->IH: The OOo font replacement will handle this. If we discover issues, we can change to other characters/fonts as well. FL->Radek: I have changed the owner to you, because you are in charge of the font.
Comment 61 nospam 2006-04-10 09:44:29 UTC
*** Issue 54450 has been marked as a duplicate of this issue. ***
Comment 62 nospam 2006-04-10 09:45:25 UTC
*** Issue 54182 has been marked as a duplicate of this issue. ***
Comment 63 mmeeks 2006-04-10 19:54:39 UTC
What remains to be done here ? - and Radek: lets get doing it :-)
Comment 64 nospam 2006-04-11 08:09:00 UTC
IH: We have to open issues here: My formula editor problem and Win98 compatibility has not been checked by now. Can anyone else also reproduce the problems with the font in the formula editor?
Comment 65 radekdoulik 2006-04-11 09:33:21 UTC
I have tested the formula editor on Linux and it works OK. IH, could you please test earlier versions of the font on windows (the one which don't have metric changes/linespacing fix suggested by cloph)?
Comment 66 nospam 2006-04-11 10:33:14 UTC
IH->Radek: Ah, this was a good hint - formula problem must have something to do with the line spacing metrics - I tested it with the font from Jan, the 5th and it worked fine - anyone any ideas?
Comment 67 nospam 2006-04-11 11:19:19 UTC
IH->cloph: It looks like we have a problem here with your new values for "Win Ascent Offset" and "Win Descent Offset". You said in #54450# that your values did not affect the placement of the brackets and other symbols. Did you test this on Win oder Linux? We have a problem on Win here. Does anyone know about the different handling of those values on Win and Linux?
Comment 68 frank.loehmann 2006-04-11 13:37:30 UTC
*** Issue 64250 has been marked as a duplicate of this issue. ***
Comment 69 lohmaier 2006-04-11 15:10:15 UTC
@ih I myself did tests on linux, - I asked members of the german community to check on windows and they reported that there were no bad sideeffects in fomrulas either. Will ask them to check with the font from this issue again. Maybe the newly added symbols have other metrics that influence the "base for the offset" (i.e. either one would have to use non-offset values or adapt these values to match the current set of symbols). In any case I would like to have the linespacing symbol resolved before replacing opensymbol.
Comment 70 lohmaier 2006-04-11 15:21:22 UTC
had a look at radek's font. It has other additional settings in the metrics tabs compared to the original font (i.e. "Typo Ascent Offset" is set to 2118 (original: 0) "Typo Descent Offset" is set to 70 (original: 0) and "HHead Descent Offset" is set to 0 (original : 471)) I'm pretty sure that this is the cause for the broken display on windows. I don't know why the other values have been changed. If there was a reason to do so, then the offset values have to be changed...
Comment 71 nospam 2006-04-12 09:08:56 UTC
IH: There is indeed a problem with the Line Spacing of the latest OpenSymbol font. If you type normal text in Writer and apply Numbering/Bullets formatting to this area you get a distance inbetween the different lines of text, but line spacing is set to single line. This only happens with this version of OpenSymbol. If you install older OpenSymbol or StarSymbol fonts everything is ok.
Comment 72 nospam 2006-04-12 09:54:55 UTC
IH: I have read the TrueType Font Spec regarding to WinAsc, WinDes, TypoAsc, TypoDes and Asc and Des in the hhea area. I have used different values for this in Fontforge and created the font, but I think Fontforge does something wrong there: my final font never containted the values I have put in the dialog in FontForge. I get the best results if I set all of those fields to Zero (standard). If I do this I have no formula problem and no line spacing problem in Writer. But I am not sure what we should do because of #54450# ?
Comment 73 nospam 2006-04-12 10:27:45 UTC
IH: I did some investigation in the internet and people there say that it depends on the application which metric values are used. So as we now know (from HDU's comment in #54450) that our Office uses the OS2 table I think it is the best idea to set the metrics in the OS2 table to the same values as in the other metrics tables in the font.
Comment 74 nospam 2006-04-12 10:40:58 UTC
IH: I have created a font with TypoAsc and WinAsc set to 1638 and TypoDes and WinDes set to 410 (not as offset values) - these values are the same as in the general tab. I get good results for the line spacing in Writer and Formula Editor. I will attach my version so someone else can confirm this.
Comment 75 nospam 2006-04-12 10:42:43 UTC
Created attachment 35654 [details] New Version of OpenSymbol with metrics set to 1638,410
Comment 76 email@example.com 2006-04-12 11:05:37 UTC
I want to remind everyone, that there already a lot of documents for which at least part of the layout depends on OpenSymbol/StarSymbol. E.g. enumerations may be affected when the line height of these fonts changes. Because of this the guys from the Writer team should carefully monitor this issue, so I CCed them.
Comment 77 Oliver-Rainer Wittmann 2006-04-12 11:10:57 UTC
replace CC member 'od' by 'fme'
Comment 78 nospam 2006-04-12 11:13:51 UTC
IH: This is a very good remark of HDU. I think we really should not do any experiments with the metrics here. I would rather do it like in my latest test version and set all metrics for Ascend and Descend to the same values in all areas without changing them to completely new ones, somewhere.
Comment 79 radekdoulik 2006-04-12 14:58:50 UTC
IH: So do you use same values as in old upstream opens___.ttf? Could you please also attach the .sfd file so that I can update ooo-build with it?
Comment 80 nospam 2006-04-13 08:52:09 UTC
Radek, I did unfortunately not use your .sdf as a basis for my testfont. Please use your font and create a new version with TypoAsc and WinAsc set to 1638 and TypoDes and WinDes set to 410 (not as offset values). According to HDU we should take those "old" values for the font not to crash the metrics of existing docs.
Comment 81 radekdoulik 2006-04-13 16:37:50 UTC
Created attachment 35683 [details] updated font with asc 1638 dsc 410 (not as offset)
Comment 82 radekdoulik 2006-04-13 16:39:58 UTC
Created attachment 35684 [details] source of updated font
Comment 83 radekdoulik 2006-04-13 16:41:48 UTC
IH, Tor: I have updated the font with these "old" values. Please check if it works OK on windows.
Comment 84 lohmaier 2006-04-13 19:04:14 UTC
The newly attached font shows the linespacing bug on linux again. As to compatibility argument: The linespacing-bug by itself is a compatibility bug (between linux and windows) Thus fixing it will change the look of old & new documents on linux, but it will fix the display only. It will not have impact on formulas, but only on normal text where symbols are taken from opensymbol. So I absolutely disagree that we should not experiment with the metrics - the opposite is the case. We should get them straight. Now is the time to fix them, since this one will replace the font and require big QA anyway. Not fixing this problem is like saying: "Documents on linux will not look the same on windows - we know that and we won't fix that" Remember that the "linebreak in justified paragraphs" behaviour was changed, modifying the look of old documents by naming it a bugfix. These kind of changes have a much bigger impact on compatibility than this one. (I can give you more examples where compatibility was sacrified/influenced) Again I ask why in Radek's font not only the values I mentioned were changed, but additional ones. (see my comment from Tue Apr 11 07:21:22 -0700)
Comment 85 radekdoulik 2006-04-13 20:37:56 UTC
Created attachment 35690 [details] another font with metrics required by cloph
Comment 86 radekdoulik 2006-04-13 20:48:25 UTC
cloph: I have created another version with your metrics, so others can test it. To answer your question, I changed only the values you suggested. Previous versions of the font already had the other values set. (as you can check from previous attachements) BTW, could you explain why do you want different asc/dsc for Win and Typo fields? Fonforge description is a little unclear on what these fields do. It says that usually Win asc/dsc fields are used for the line spacing on Win, so in this case it doesn't matter what is set in Typo fields. (but the description might be wrong) Anyone know how is it then? And how did you get/calculated these values?
Comment 87 lohmaier 2006-04-13 23:45:01 UTC
> cloph: I have created another version with your metrics, so others can test it. Thanks. Looks good on linux (i.e. linespacing problem solved and no other problems with formulas) > To answer your question, I changed only the values you suggested. Previous > versions of the font already had the other values set. (as you can check from > previous attachements) I did not compare the settings to the ones attached to this issue, but to the ones currently shipped with OOo/the one in cvs (md5sum dd604fd024ebb8efc7872e2f5dd4b927) The very first one attached to the issue has all metrics set to zero btw. (already different to the one in cvs), the second one marked with "created with fontforge" has the Typo-values set, so is already different from the first one and from the one in cvs.. > BTW, could you explain why do you want different asc/dsc for Win and Typo > fields? There is no specific reason. I determined the values by trial and error and the Win-Values made the difference... I didn't touch those that did not make a difference on linux. And in the "original" version, the Typo values are set to zero. As http://fontforge.sourceforge.net/editexample5.html#baseline reads the Typo-Values are used on windows, that's why I didn't touch them and want to have them reset to the values from cvs that are known to work. They may need some minor finetuning, but values of 2000+ offset just seem wrong to me. (and apparently caused problems) It also writes that Windows-programs often don't care and take the win-values instead (and since simonAW tested the values on windows and said they were OK I still think they'll don't cause the display-problem in the attached formula-doc) In http://fontforge.sourceforge.net/fontinfo.html#TTF-Metrics it already reads: "The Typographic Ascent and Typographic Descent are /supposed/ to represent the line spacing of the font on the windows platform. Sadly very few applications actually use them (most applications use the Windows Ascent/Descent described above)." > And how did you get/calculated these values? I did not calculate them. They were determined by modifying the value, installing the font, launching OOo, test whether the effect still is visible, modify it again, ....
Comment 88 radekdoulik 2006-04-14 10:15:50 UTC
> In http://fontforge.sourceforge.net/fontinfo.html#TTF-Metrics it already reads: > "The Typographic Ascent and Typographic Descent are /supposed/ to represent the > line spacing of the font on the windows platform. Sadly very few applications > actually use them (most applications use the Windows Ascent/Descent described > above)." Yeah, and that seems to be the source of this problem. The font worked OK on windows before win asc/dsc changed, so I guess the latest attached font will have the same problems on windows again. I would prefer if we use the font with original metrics as suggested by hdu. That way we will fix the bug this issue was open for. Let solve the other problems in different issues later. Trying to solve more problems at once seems to get us nothing (so far).
Comment 89 pavel 2006-05-23 06:58:26 UTC
Reassign to hdu, re-target because it is too late for 2.0.3.
Comment 90 radekdoulik 2006-05-29 13:50:19 UTC
I have created new bug for the linespacing issue. HDU, IH: could we get the font with original metrics upstream and finally close this long issue? ;-)
Comment 91 radekdoulik 2006-05-29 13:51:39 UTC
the linespacing issue: http://qa.openoffice.org/issues/show_bug.cgi?id=65855
Comment 92 nospam 2006-05-29 14:27:23 UTC
IH: good idea, radek. please create a CWS and check in your font that fixes this bug here and then we can take care of the other (line spacing) problem.
Comment 93 radekdoulik 2006-07-26 09:38:19 UTC
OK, cws created (sorry for delay). It is named opensymbol01 and contains the fixed font and the font source. Not sure how to proceed now, how do I ask for QA to start?
Comment 94 radekdoulik 2006-07-26 09:41:36 UTC
marking FIXED as it landed in the cws
Comment 95 radekdoulik 2006-07-26 09:44:37 UTC
IH: not sure whether you want to fix the linespacing problem in that cws or later in other. I would like to set cws status to "ready for QA", is that OK with you?
Comment 96 nospam 2006-07-26 10:10:20 UTC
IH->Radek: it is ok for me - fix this step by step and do the linespacing issue in another one - so you can set this cws to ready for qa - please set the QA-Representative for this cws to mru and send this bug to mru for testing. Thank you, Radek for your good work.
Comment 97 nospam 2006-07-26 10:32:29 UTC
IH->Radek: I've talked to mru from our QA and this is also ok for him. He we also take care that this cws gets the right release target flag for 2.0.4.
Comment 98 michael.ruess 2006-08-03 14:04:30 UTC
Retargeted to next release due to resource limitations.
Comment 99 nospam 2006-08-09 09:44:06 UTC
IH: It is really no good news that Radek's fix will not make it into 2.0.4. I am very disappointed - this has always been postponed since 2.0.2
Comment 100 michael.ruess 2006-08-09 14:23:11 UTC
Targetting issue to newly introduced "OOo 2.1".
Comment 101 michael.ruess 2006-08-14 13:37:09 UTC
*** Issue 67782 has been marked as a duplicate of this issue. ***
Comment 102 michael.ruess 2006-09-01 14:07:21 UTC
I have now tested this font extensively on Win and Unix platforms and in my eyes it looks VERY good!!! Maybe somewhone can pst somewhere a message, tat a new version could be downloaded from this issue. Thanks for the good work on it, Radek!
Comment 103 Regina Henschel 2006-09-03 13:41:30 UTC
In the version "Apr 13 12:37:00" the upper part of the characters is missing, not only in OOo but also in WordPad. (Win XP Home SP2)
Comment 104 michael.ruess 2006-09-04 14:58:08 UTC
I cannot share Regina's notice. I tested on Win98/2000/XP, Linux and Solaris. The font from mentioned date looks good (no missing characters in the top part). Maybe you could attach a screenshot of the "Insert/Special character" dialog with the part displaying wrongly on your system?
Comment 105 Regina Henschel 2006-09-04 15:39:31 UTC
Created attachment 38939 [details] Screenshot from OOobuild9062; WinXp desktop 1024x768 Pixel;GroÃŸ 120DPI; Clear Type
Comment 106 Regina Henschel 2006-09-04 15:44:03 UTC
Screenshot of textdocument attached, insert special dialog will follow. The version "Apr 13 08:37:00" is OK for me.
Comment 107 Regina Henschel 2006-09-04 15:45:07 UTC
Created attachment 38940 [details] sreenshot of insert special character dialog
Comment 108 radekdoulik 2006-09-04 15:49:34 UTC
regina, which font do you use? There were 2 fonts added on Apr 13th. The right one should be: Thu Apr 13 08:37:00 -0700 2006: opens___.ttf updated font with asc 1638 dsc 410 (not as offset) (application/octet-stream) http://www.openoffice.org/nonav/issues/showattachment.cgi/35683/opens___.ttf
Comment 109 radekdoulik 2006-09-04 15:52:47 UTC
OKie, you answered when I was still typing the question :) Yeah, the "morning" version is the right one.
Comment 110 caolanm 2006-11-04 14:39:22 UTC
this can be closed now for 2.1, right ?
Comment 111 michael.ruess 2006-11-24 11:24:46 UTC
Yeah, checked it in 680m188 OOo build. CLosing issue.