search

praveenbankbazaar / httparchive

Automatically exported from code.google.com/p/httparchive
0 stars 0 forks source link

Proposed Correlations Table #371

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
This is a suggestion rather than a defect but recently, when I was working with 
the new changes, I found it a little difficult to understand the correlation 
charts. My suggestion would be to break the correlations out of the statistics 
table and into a separate table. This would bring both more flexibility - you 
can easily vary the number of terms used per run - and clarity as the labelling 
can be handled in the table. The schema would be something like the following, 
note I've added FOREIGN KEYS for completeness.

CREATE TABLE `http`.`correlations` (
  `label` VARCHAR(32) NOT NULL COMMENT 'The label of relevant run.',
  `device` VARCHAR(32) NOT NULL COMMENT 'The device of the relevant run',
  `category` VARCHAR(32) NOT NULL COMMENT 'What the value correlates with.',
  `term` VARCHAR(32) NOT NULL COMMENT 'Description of the statistic',
  `value` DECIMAL NOT NULL COMMENT 'The degree of correlation',
  PRIMARY KEY (`label`, `device`, `category`, `term`),
  CONSTRAINT `label` FOREIGN KEY `label` (`label`)
    REFERENCES `stats` (`label`),
  CONSTRAINT `device` FOREIGN KEY `device` (`device`)
    REFERENCES `stats` (`device`)
)

COMMENT = 'Correlations between particular statistics and page loading';

Original issue reported on code.google.com by charlie....@clark-consulting.eu on 4 Mar 2013 at 4:55

GoogleCodeExporter commented 9 years ago 
Not sure what the desired outcome is. If the goal is a table that is more 
understandable for httparchive developers, then I don't think the benefits 
outweigh the costs. If the goal is a better looking chart, please explain how 
the chart improves. Marking "wontfix" for now until clarification.

Original comment by stevesou...@gmail.com on 10 May 2014 at 4:13