Converting CVSNT to Linux CVS
Today at work we swapped from using CVSNT on a Windows server machine to using CVS for linux on a QNAP NAS storage box. All appeared to be working fine until I checked out a file that was committed as a binary file. I quickly realised that half of the repository was corrupt (i.e. all binary files). After a little bit of research it was clear that CVSNT and CVS for linux are not directly compatible any more as they store the fact that a file is binary in a different format.
In each repository file there are entries that specify file formats, cvs comments, versions, etc. In amongst these is a flag that specifies the file type.
A file specified as binary in CVSNT has a line reading (where the space is actually a tab):
kopt b;
A file specified as binary in CVS has a line reading (where the space is actually a tab):
expand @b@;
Once I realised this was the case it was a fairly simple matter of converting these line in all files in the repository. To do this I used a usefull little application I have used before called 'TextCrawler'. This is a free download and is useful for performing find/replace actions on multiple files in a directory.
Before I performed the correction I ensured that all files were marked as writable (ie unchecked read-only flag for this directory and applied to all sub files and folders).
Download text crawler and set the options shown in the image below:
Clicking 'Replace' will now search through all files in your repository (my test location here is c:\CVSDATA\).
Once the operation is complete the files are ready to be used in the linux CVS repository.
