Brad Chapman wrote: > > I'm using the latest version (1.5.2). I took a look at the lib > documentation for os.makedirs and found the following: [...] > The default mode is 0777 (octal). New in version 1.5.2. That's the problem. I'm using 1.5.1, which is standard with RH6.0. > So it appears as if this will require 1.5.2 to run. My setup for python is > the default so hopefully the problem should only be a 1.5.2 versus older > versions problem. I know there was mention of a RedHat issue with 1.5.2 > versus 1.5 on the omniORBpy mailing list. I think some versions of RedHat > ship with 1.5 by default, so sometimes even if people had 1.5.2 installed, > 1.5 would still be in their path. Does RedHat 6.1 also come with 1.5.1? > If it is a problem to use 1.5.2 only stuff, let me know and I'll > think of a workaround here. I like os.makedirs() over os.mkdir() because > the former will put in all intermediate directories when making > directories, but I can do something different. 1.5.2 is still pretty new, so I don't know how many users will have it. Do we want to make some 95% of our users upgrade their Python installation? What's the concensus here? Who on this list wants to stick with Python 1.5.1? (I may have to send this as a separate message to get some people's attention.) I think we could have Loci check the version of Python being used and select a method accordingly. Or maybe we could hold off on os.makedirs() until 1.5.2 becomes more common. Cheers. Jeff -- +----------------------------------+ | J.W. Bizzaro | | | | http://bioinformatics.org/~jeff/ | | | | THE OPEN LAB | | Open Source Bioinformatics | | | | http://bioinformatics.org/ | +----------------------------------+