Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.
On Tue, 4 Mar 2003 11:10:18 -0800, Ian Epperson <Ian at axiomdesign.com> wrote: >The other day, something frightening happened. > >I was working remotely (sick - working from my home) coordinating program >changes with programmers at another remote site (client's home). I made 4 >small changes to 3 different files, and committed those three files. The >other site updated, compiled, and didn't see all my changes. I looked, and >the 3rd file, though committed, had my change missing. I ran the CVS >History command, and saw my commit. I did a diff between that version and >the previous version and the only differences were in the CVS keywords >($Date$ and $Author$) with only one of my two changes (to this file). A >routine that I had just written was simply gone. I did a find for that >routine (to make sure I didn't add it to the wrong file), but couldn't find >it in ANY file. > We used to get complaints of that ages ago, and it was always down to someone leaving an IDE open when using CVS. The developers are now trained to shut the IDE down when committing and updating and nothing has been lost since AFAIK (and I would probably hear about it if it was... I get earache every time anything doesn't work quite how they expect). It's possible that someone committed a delete of the routine you just wrote, or even played fast and loose with the 'admin -o' command (unlikely I expect). The diff code hasn't changed much since the early betas - I try to leave it alone since it's so critical to the operation of CVS. I'd be surprised if it was anything inherent in GNU diff that was causing the problem. Tony