|
From: Colin S. <col...@ex...> - 2004-07-19 15:41:48
|
CVS will convert end of lines properly on checkin and checkout, appropriate to the platform being checked out on. However, it is unable to handle the case where somebody creates or modifies a file on Windows (which has CR/LF at the end of lines), then mails it (or whtever) to somebody who then checks it in from a unix system. It will then appear in CVS with the CR still there since that system will not strip out the CR. Then when you subsequently check it out on the windows system, CVS will add another CR, so you end up with CRCRLF. Colin Darren Davison wrote: >>Hi, >> >> >> >>>- We need to adapt the source code formatting of the JMS >>>support: I still see nasty empty lines after each code line >>>on Windows: maybe Unix/Windows line feed differences? >>> >>> >>I'm not seeing this, if you show me how you are detecting this I can clean >>it up. >> >> > >The JMS files are using windows EOL markers (0A 0D) the other files use >the Unix convention (0A). You can verify by opening the files with any >binary viewer (TextPad on Windows will do it, any Hex viewer on Unix/Linux >will show it too). I thought the CVS server should automatically convert >to the Unix format..? > >I downloaded AbstractJmsTemplate.java and checked it with both TextPad on >'doze and the nano editor (pico clone) on Linux; they formatted OK on both >for me. Emacs would probably show the JMS source as having ^M at the end >of each line. > > > > |