Community
Participate
Working Groups
The list of mirror sites is not displayed correctly because of the RTL reading order. Since this list will always contain English strings, it should be set with LTR reading order (Not the dialog title or Hebrew description text though). This article was reassigned from Category:''Legacy Project,TVT,Uncategorized''.
Created attachment 31565 [details] site_mirrors_rtl.jpg
<cde:tctdetail> Testcase: 09.003020 Project: WSW3A Component: Platform Priority: 2 Subject: List of update site mirrors should be LTR Article ID: 243 Originator: gpelleg@il.ibm.com </cde:tctdetail>
Dejan, do you know where the list of mirror sites comes from? In order to resolve this bug we would have to have the mirror site names in a properties file so the RTL language translators could insert LRM markers into the string to display it correctly in RTL orientation. I am thinking we do not have access to these strings to alter them in such a way - is that correct?
For 3.1.*, the list of mirrors comes from a file listed in site.xml. Last time I checked, this file was owned by either Releng or Eclipse webmaster.
Sonia, is the list of mirror sites owned by releng?
This is owned by webmaster@eclipse.org. There is a lot of work going on in this area right now (bug 99412).
LRM markers could be inserted into the text after the closing ]. For more info on this problem see bug 119927, bug 119517 or contact me (or Tod if I am not available).
Can someone give me a hint on what I need to do to fix this? D.
Are the update sites just strings in a file that are read by the update code in Eclipse at some point? Perhaps it would be better if the update component were to take the strings it gets from the update site files and run it through the TextProcessor.process(String, String) API. This will work if the set of neutral characters is definitive - neutral characters are: . , ; ! ? ~ @ # $ % ^ & * ( ) { } [ ] < > \ / " ' ` | If that is not the case, then you would have to insert LRM(\u200e), PDF(\u202c), and LRE(\u202a) markers into the text file(s) that list the update sites. The basic algorithm is: 1. Scan the string to locate the separators and the tokens. 2. While scanning, note the direction of the last strong character scanned. Strong characters are characters which have a BiDi classification of L, R or AL as defined in the Unicode standard. 3. If the last strong character before a separator is of class R or AL, add a LRM before the separator. Since LRM itself is a strong L character, following separators do not need an LRM until a strong R or AL character is found. 4.If the component where the pattern is displayed has a RTL basic direction, add a LRE at the beginning of the pattern and a PDF at its end.
(In reply to comment #9) > Are the update sites just strings in a file that are read by the update code in > Eclipse at some point? Yes. Update reads an XML file on the server. > Perhaps it would be better if the update component were > to take the strings it gets from the update site files and run it through the > TextProcessor.process(String, String) API. This would likely be a solution that would provide a wider benefit. I can change the [ ] to ( ) if that helps, or use Country - mirror name instead. Let me know what I should do. D.
Target for Q4
I created a Label and then a List and used a string that resembled one of the mirror sites. If I use a Shell to show either of these widgets the string renders properly. If I use a JFace Dialog (which is what the MirrorsDialog is), the string no longer renders properly. Bug 124517 could be related. I think you're off the hook for this one.
Susan, it appears something in the JFace layer is causing this problem. bug 124517 may be related.
changing prio and status to conform to plat-ui bug guidelines
This dialog is no longer supported (Update manager has been replaced by p2).
