Stefan Sperling <s...@elego.de> writes: > On Mon, Jun 25, 2012 at 05:21:53PM +0200, Bert Huijben wrote: >> (I don't see how it can corrupt your working copy. It can make a local >> change unnoticed, but I wouldn't call that corrupted) >> >> Bert > > I just meant to say that the db state is inconsistent with the > expected state if this bug triggers. If that is not corruption, > then we can label it "inconsistency" or whatever.
I'm trying to identify what exactly goes wrong. The main way to trigger the bug is "svn propdel svn:eol-style" when svn:keywords is not set. The result is that translated_size remains set when it should be -1. I think this could only be a problem when svn:eol-style changes the size of the file, i.e. svn:eol-style=native on Windows, but I'm still not sure what effect the bug will have. A pristine file with content "fooLF" and svn:eol-style=native will have a working file "fooCRLF" and translated_size=5 on Windows. If we delete the svn:eol-style what goes wrong? I suppose the working file should get converted to "fooLF" on commit. Does this fail to occur? -- Cerified & Supported Apache Subversion Downloads: http://www.wandisco.com/subversion/download
