Dear all, I have a wierd bug, which I reported to this mailing list earlier, but there was no response. I am posting it again, with some additional info:
I am using OOo 1.9.72 I have some text in a CTL language (Malayalam) in an odt file. I also have two Malayalam fonts, which have all the necessary OT tables for proper Malayalam shaping. I have verified this independently of OOo. Let us name it font A and font B. Experiment 1: When I open the file, the text doesnt show correctly: it is as if each Unicode codepoint is rendered one after another, instead of the appropriate shaping for Malayalam. However, if I change the font to another Malayalam font (font B), the text renders OK. I close the file without saving. I open the file again, and this time the text shows correctly, even without changing the font. I close the file. I open the file the third time, the text doesnt show correctly. Changing the font to another (font B), it shows correctly. I close the file without saving. Opening and then closing the file, and again opening it, switches the rendering. Alternatively the text renders incorrectly and then correctly. Experiment 2: Same as experiment 1, but this time, when the text shows correctly in font A, changing it to font B, shows the *incorrect* rendering. This alternation is also consistent. Experiment 3: My OOo setting is to use the system fonts for the display. This is so that, I can see shaped Malayalam in the UI elements, for e.g., in the "Input Line" widget in the Calc application. I use the font B for UI. I repeat Experiment 1 and 2. Whenever the text renders *incorrectly*, the UI text renders correctly, and vice versa. This occurs consistently. Experiment 4: I thought perhaps the shaping engine code wasnt being called. So I added debug statements to ICU layout engine code. The code is called and tables evaluated *everytime*, whether the text renders or not. Seeing the 4 experiments above, can *anyone* suggest something that I could do to figure out the error ? This completely eludes me. What can I do to find the point of error ? Rajeev J Sebastian --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
