database

[tealover]

Hello,
I've got two questions I couldn't resolve myself so far:
1. how would I COPY one of the databases? I want to modify values,
that's why I want a copy, so VIEW is no good here.
2. Where exactly is "per developer" computation done?
3. And, at which point timestamps (from Revision table) are converted
into days for display purposes?

Thank you so much!

[David James]

On Apr 4, 2005 1:36 AM, tealover <teal...@yandex.ru> wrote:
> 2. Where exactly is "per developer" computation done?

That'd be ca.utoronto.JCVSReport.report.BasicGraph. Search for
"perDeveloper" in that file.

> 3. And, at which point timestamps (from Revision table) are converted
> into days for display purposes?

That's in ca.utoronto.JCVSReport.metric.MetricRevision.getDay.

> 1. how would I COPY one of the databases? I want to modify values,
> that's why I want a copy, so VIEW is no good here.

I recommend you ask this question on the CSC408 bulletin board or on
an SQL-related forum.

Best,

David
--
http://www.cs.toronto.edu/~james

[tealover]

Another question:
since McKoi database doesn't support any sql functions beyond basic
stuff, I cannot just just write a specific querry to get the info and
will have to modify the code to treat the special case of the new
metric, so the question is then where would be the least painful way to
insert a clause "if the case of new metric, process db table in the
follwoing way..."

Thanks a lot in advance

[tealover]

Yet another question:
how does incremental stats collection work for, say, number of commits
metric?
I would have assumed that collecting this data incrementally would mean
that values grow or stay the same.
It is not the case in the graphs, so I'm wondering how it works.

Thank you very much,
Irina

[David James]

Hi Irina,

Nice to hear from you. Hope your project is going well.

Looks like there's been quite a few questions, more than I can keep up
with! I'll try to answer your questions, but understand I'm a bit
swamped because I've got four assignments due Friday. Ah, the life of
a student!

> how does incremental stats collection work for, say, number of commits
> metric?
> I would have assumed that collecting this data incrementally would mean
> that values grow or stay the same.

That's exactly how it works. If the graph is marked as 'incremental',
then for each new datapoint, we report the sum of all the datapoints
gathered so far.

Take a look at one of the CVS Operations graph, for example:
http://www.cs.toronto.edu/~james/408/Operations.jpg

> where would be the least painful way to
> insert a clause "if the case of new metric, process db table in the
> follwoing way..."

You're asking how to change the way that metrics are processed after
they have been added to the database? That's actually quite easy --
not painful at all! :)

Let's say you wrote a class for processing metrics called
'PerDeveloperScatterPlot', following the same convention as the other
classes in the Report package such as PerDeveloperTimeSeries.

If you want to display any metric using your new class, simply specify
as such in your sample_config.txt. Instead of specifying simply
'CodeSize-PerDeveloper', specify 'CodeSize-PerDeveloperScatterPlot'.
Then, instead of displaying the number of classes as a per developer
pie graph, you'll display it as a scatter plot!

Hope that helps!