Libreoffice Writer - Some marks in the Unicode font shows dotted circle

I am using Libreoffice Writer in Linux Mint. I have a unicode font installed in Linux Mint. There are three marks (diacritics) in this font which I often use. Only one of the marks renders well in Libreoffice Writer where as the remaining two marks are always rendered with dotted circle. When I use other apps in operating system, such as xed (text editor in Linux Mint), all the marks render well.

Please see below two images. First image is Libreoffice Writer where only one of the marks renders well (highlighted in green) and remaining two marks rendering with dotted circle (highlighted in yellow). Second image is xed text editor in the same operating system which correctly renders all marks.

Any help to achieve proper rendering in Libreoffice Writer is highly appreciated.

The font works well in Windows and Mac with Microsoft office. It works well with text editors in Linux. But it is not rendering well some marks in Libreoffice Writer in Linux, Windows, Mac - the issue is with Writer in all three operating systems.

Sample Doc With Font Issue.odt (14.7 KB)

Dotted circle display like this means the Unicode sequence is not valid, i.e. the diacritics could not be applied to a non-diacritic glyph. This may be a typing order problem. To check if there is a bug in the font renderer engine or somewhere else, attach the .odt file so that we can have access to the raw byte sequence.

1 Like

Thank you for responding. I have attached the .odt file with sample text. I have also highlighted in yellow those locations where the rendering is not correct. As it is evident from the document, the same mark (vertical line above) sometimes renders correctly and sometimes not. Same issue with other mark (horizontal line below) which also sometimes renders correctly and sometimes not.

Same paragraph of text when copied into another text editor in Linux and when applied with the font that I am using, all marks are rendered well.

This looks like a bug; the DEVANAGARI STRESS SIGN UDATTA is shown separately from the preceding TELUGU LETTER MA, when using Mandali, which has both. Possibly a reason to file a bug report; @khaledhosny2 could know better.

1 Like

Looking at మ॑ (first yellow highlighted sequence)
It is made of U+0C2E (TELEGU LETTER MA) and U+0951 (DEVANAGARI STRESS SIGN UDATTA). Putting aside the fact that you mix two scripts, U+0951 is defined as a non spacing mark. I don’t know what you expect layout-wise (because I am not at all familiar with India scripts) but displaying it immediately at right of base letter seems logical.

However, after I enable Telugu in Tools>Options, Language Settings>Languages, UDATTA has a slightly changed shape and is shifted a bit leftward above the last “loop” of preceding letter.

This activation changes display of your highlighted areas. Since I don’t know what to look for, here is my screen shot:

Is this better?

In which case, check you have enabled Complex text layout for the correct language in Tools>Options, Language Settings>Languages

I think the .odt document did not embed the font with which I am facing this issue. Here is the screenshot of .odt file contents showing the issue (highlighted locations are those where mark rendered with dotted circles).

I have the “Complex text layout” box checked.

You’re right. Mandali Modified shows up italic in the font menu. Consequently a substitute was used, but I don’t think font should matter unless the glyphs have not been assigned the “combining” attribute by font designer.

Note that U+0951 is defined as “used also in Vedic texts written in other scripts”, so no problem here.

IMO, this is a bug. The settings should not change any rendering, and should only change the UI.

Yet it does :wink:

The issue is with DEVANAGARI STRESS SIGN UDATTA and also with DEVANAGARI STRESS SIGN ANUDATTA. Whenever these two unicode characters are combined with Telugu language letters they are not rendering properly. The font I am using contains both the Devanagari signs as well as Telugu characters. The font itself renders the combination of these devanagari signs with Telugu characters very well in other application within the Linux (and also in MS Office apps within Windows, Mac). The issue appears to be not related to font but I guess it is the font when used in Libreoffice Writer (I am not an expert though)

Here is what it looks like initially without enabling Telugu script:

The font itself works well in other apps within Linux (and also with MS office apps within Windows and Macos). Whereas the font has same issue with Libreoffice writer in all these operating systems. Therefore I am not sure that the issue is with font itself.

As we can see in the document, I have also used VEDIC TONE DOUBLE SVARITA in combination with Telugu characters. My observation is, every time I use this sign in combination with any Telugu character, it renders perfectly. I have checked the font itself to understand how differently the double svarita was designed in the font compared to udatta and anudatta. I do not see any difference across these three marks within the font design.

The issue is therefore, udatta and anudatta when combined with Telugu characters end up with dotted circle. Whereas double svarita when combined with Telugu characters always renders correctly. However the font itself works well in other apps (all three i.e. udatta, anudatta, double svarita in combination with Telugu characters renders well).

I am unsure what you want to point to by this; in any case, the combining in LibreOffice definitely depends on the font, as shown by @ajlittoz using some unknown substituted font, which combines the glyphs. This doesn’t mean that the font is definitely “problematic”, but it shows that when you follow my advise, and file the bug, you definitely must mention the font, and provide it. LibreOffice may have problems with some aspects of valid fonts, or the font may really have some issues - only experts could tell.

I have filed Bug 155514

I had a look to Mandali with FontForge. The Devanagari block contains only 5 glyphs and indeed UDATTA and ANUDATTA are encoded. However, when I compare to common Western European fonts, these diacritics are flagged “Base Glyph” while diacritics are “No Class” in other fonts. Don’t know if this is relevant.

Note also Mandali is © 2012, which is eternity with computers.