Re: Your gcc3.1 release date is a joke

[Tim Josling <tej at melbpc dot org dot au>]

The scheduled date in IT projects is usually defined as 'the earliest date which cannot be proven to be impossible'. Optimistic estimates are a chronic problem in GCC as elsewhere.

While the tone of the email was immature and offensive, he does have a point. But it is more about setting expectations than how long things are actually taking.

My cobol compiler is running a few years late, though I have had no complaints, perhaps due to my being very careful not to create expectations I cannot meet.

We do not have a PHB creating artificial deadlines, so we can set our own, realistic schedules and not create expectations that are unrealistic.

Maybe we can learn from this for 3.2. Why not take as a starting point 'how long gcc3.1 took' and add a tolerance of +/-50%. Maybe things will improve, but usually, expectations of a massive
improvement are not realised.

So we could say something like:

"GCC 3.2 final is expected to be completed somewhere between Dec 2003 and September 2004. Please note that this is an expectation, not a commitment, and the dates may change if this is necessary to
meet quality standards or if the level of support is less than expected. You can help move this date forward by providing lawyers, guns, code and/or money. You can use pre release versions if you
require GCC 3.2 features. As always however you should test software to ensure it meets your needs. "

Tim Josling