|
Apache OpenOffice (AOO) Bugzilla – Full Text Issue Listing |
| Summary: | basic/inc/testtool.hxx #defines conflict with basic/inc/sbdef.hxx enum | ||||||
|---|---|---|---|---|---|---|---|
| Product: | General | Reporter: | fa <dan> | ||||
| Component: | code | Assignee: | fa <dan> | ||||
| Status: | CLOSED FIXED | QA Contact: | issues@framework <issues> | ||||
| Severity: | Trivial | ||||||
| Priority: | P3 | CC: | caolanm, gregor.hartmann, issues, jens-heiner.rechtien, nesshof | ||||
| Version: | 680 | ||||||
| Target Milestone: | OOo 2.0 | ||||||
| Hardware: | All | ||||||
| OS: | All | ||||||
| Issue Type: | DEFECT | Latest Confirmation in: | --- | ||||
| Developer Difficulty: | --- | ||||||
| Issue Depends on: | |||||||
| Issue Blocks: | 28294 | ||||||
| Attachments: |
|
||||||
target -> 2.0 Created attachment 16465 [details]
Increase explicit size of enum to cover all values
Does the patch seem correct to everyone? It fixes the problem, but I still think the correct solution is to not have the extra TT_* values and cast them to SbTextType. of: Please have a look. Is the target ok? reassign back ;) Committed to CWS gcc340fixes01 HR->GH: Dans patch is correct and will do the job, but this really needs a thorough cleanup. of @ mh: Please enlighten me: Why is this 'component framework'? What has this to do with the office framework? mh->of: in this case it is in fact not easy. The basic project is located in the CVS in the Script project which was declared as "dead". So usually you take the CVS Project as corresponding component in IZ, and for basic in this case we take framework as an exception. In every case this is not an porting issue but has to be addessed by the developer directly. In this special case the best way would have been that Dan already had assigned this issues to himself at submit time, so we both wouldn't have been involved at all :). of @ mh: Got it. :-) I wanted clarification from the original Sun developer or someone familiar with the code to tell me why the SbTextType was getting used in this way... I did not feel completely comfortable with the patch and wanted to make sure the original intent was correct, and that this problem was not somehow a coding mistake or mis-design. closing ... |
gcc 3.4 does not allow an enum type to be assigned a value outside its declared range. In sbdef.hxx, we have: enum SbTextType { SB_KEYWORD = 1, SB_SYMBOL, SB_STRING, SB_NUMBER, SB_PUNCTUATION, SB_COMMENT }; In testtool.hxx, additional values are defined, but using #define and they are not part of the enum. This is illegal in C++ with gcc 3.4: #define TT_KEYWORD ((SbTextType)100) #define TT_REMOTECMD ((SbTextType)101) I've looked over the sources, and it seems that the testtool.hxx types are not widely used (used only in 3 of 4 files). I don't think the "clean" solution is to add the testtool.hxx ones to the SbTextTypes enum because they are clearly not "normal" code. However, I'd like insight into how to fix this problem. Any ideas?