[unixODBC-dev] what should I fix?

Nick Gorham nick at lurcher.org
Mon Mar 7 19:32:26 GMT 2005

Eric Sharkey wrote:

>>Well, my position, is that any change that goes into the directories I 
>>mentioned, I would like to look over before the change gets forgoten 
>>about, as they are the bits that will cause a lot of people a lot of 
>>pain if they cause problems that are not caught before the next release.
>You seem to be dancing around the question I'm trying to ask.
>Let me rephrase.
>By what method do you prefer to look at these changes?:
>  1. Patch files generated by the author and mailed to you/this list
>     before commission
>  2. Patch files generated by you via "cvs diff" commands, after the
>     author has placed the patch in the repository
>Some people like tight control of their sources and insist on method
>1.  Frequently they feel that this improves reliability and security.
>In my experience, method 2 is actually the more secure and reliable
>method for collaborative development among untrusted developers,
>since what you're looking at is the actual source code change, and there
>may be small differences between posted and actual patches, either
>intentionally or by error.
Ok, maybe I am not answering the correct question.

I was under the impression that the cvs mailing list would let me see 
any changes that were commited to the tree. Thats enough for me, I can 
take a look at them and if they look like I would have a opinion then I 
can just ask and we can discuss

All I was suggesting about the list of directories I was worried about, 
was I would prefer discussing changes to those before the changes 
started. Once the change discussed was agreed, then if it wasn't me 
making the change, I would check it out after it was commited.

I thought your main area of work was going to be on the testing parts of 
the distrib and as such there is little "prior art" to effect. If the 
tests find problems in the DM and those "gold" bits, I am quite happy to 
fix them, assuming the tests can show the problem, and we all agree that 
its a real problem.

Part of the reason I am wary of changing the DM code isn't that I don't 
think its got bugs, all software has, but I know that the bugs that are 
there don;t show up on the bunch of tests IBM rad against it, so I don't 
think they are big killer bugs.


More information about the unixODBC-dev mailing list