Age | Commit message (Collapse) | Author |
|
This makes things a little nice to compare in relation to each other.
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
latest version includes:
* removal "toolchain" left over from last release
* enhancement for index page to support comments
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
This is a small hack to allow providing comments to the "index"
report we make. It queries the database for a report named "index"
if this exists, it will use its comments when rendering the report.
NOTE: The "Can publish" should be left unset (false) so that it doesn't
show up as a normal report users can view.
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
This was left out in a prior commit
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
Doing the query by test doesn't return the proper metrics. We
instead need to query by the actual test name.
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
These reports are really just system-wide benchmarks and are
not 100% dependent on the toolchain.
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
The GarbageCollection test in 0xbench had two issues:
1) its measurement size is a couple of orders of magnitude bigger than
than the other measurements, so the graphs look bad.
2) Smaller results are better
This breaks out that test into its own run so that we can
better display results to the user.
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
This adds a new field "can_publish" to a report. Until this
gets set to True, it won't get listed in the main view. This
means only users with the actual URL to the report can view it.
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
The bundle drop-down box in the admin panel was way too
big when taking into account all the bundles installed on
validation.linaro.org. This updates the admin interface to
take the bundle ID instead.
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
The old calculation was wrong and only worked with 3
benchmark runs.
This spells out the math more cleanly and has been tested
for with 2,3,4 and 5 benchmark runs to ensure it scales properly.
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
It really only needs to allow for an empty entry. I'm also
adding some help text to the field to describe its usage, since
the name comments could be assumed to be something else.
Signed-off-by: Andy Doan <andy.doan@linaro.org>
Change-Id: Ibd6569a97121c0c00c47b77175873a24d4dc1db1
|
|
dashboard_app templates are not guaranteed to be a stable interface
Signed-off-by: Andy Doan <andy.doan@linaro.org>
Change-Id: If208a06bbbba58bbdc37703ac9d281074586e10f
|
|
Remove an unused function
Add comments to help describe what the model objects do
Signed-off-by: Andy Doan <andy.doan@linaro.org>
Change-Id: I8dc152408090fdd53a93c32c8ea0bb8fae985b03
|
|
This enables support for the permalink system
Signed-off-by: Andy Doan <andy.doan@linaro.org>
Change-Id: Ie1f64f8f0a2068269e39546cd07de8c8f6ebc2cc
|
|
This provides the ability for reverse lookups.
Signed-off-by: Andy Doan <andy.doan@linaro.org>
Change-Id: I6efe574bd1da4add09451c90cd2537c7ab2c9d60
|
|
Signed-off-by: Andy Doan <andy.doan@linaro.org>
Change-Id: Ic2703195135862e0c08d81912229cf7376fd5654
|
|
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
Signed-off-by: Andy Doan <andy.doan@linaro.org>
|
|
This changes the main page to show a historical view
of linaro-4.6 performance for the last 6 months. It tricks
out the report.html framework that already exists in order
to produce the view
|
|
This removes the hard-coding of the chart width. It also changes
the direction of the text. When the chart was 100% width, it
was causing the labels to overflow on the right hand side of
the page.
|
|
The main support for this comes from:
http://people.iola.dk/olau/flot/jquery.flot.navigate.js
|
|
|
|
This allows the admin to provide some comments about the
report. For example, an executive summary
|
|
If the test order isn't the same in each benchmark run
things go haywire because we assume the arrays are in the
same order. This changes the code to use the order of things
in the first run to be the reference order and then we search
for results in other runs based on that order.
|
|
This moves all the code away from the original code
that messed around with the JSON representation of the
report and puts the data together using the queries from
the BenchmarkRuns and then assembles them using JavaScript.
|
|
This helps move off the benchmark_run.py in order to use
the django model objects more efficiently. Functions in this
module are being transitioned to helpers.py to more closely
follow the django pattern used by the LAVA dashboard project
|
|
|
|
the angled text makes labels take up more room than originally done.
this gives us a little more room for the actual chart view
|
|
The flot code applies the error amount to both above and
below the average. This reduces the standard deviation in
half so the error range is the proper size
|
|
The order can be important. For instance the last test in some
benchmarks is the overall total
|
|
the current logic didn't support multi-bar graphs and charted
the error on the edge of a bar. This caused error bars to get
overlapped and made it impossible to read things. The new change
offset the error bar to the center of its bar
|
|
based on patches from:
http://code.google.com/p/flot/issues/detail?id=159
|
|
based on patch from:
http://code.google.com/p/flot/issues/attachmentText?id=215&aid=3648196345862516676&name=jquery.flot.errorbars.js&token=e8fc53f7cf176b0a1961ea9aece25e64
|
|
based on patch from:
http://code.google.com/p/flot/issues/attachmentText?id=85&aid=-5081524213585119462&name=angledlabels.patch&token=p1H3bJn7coHFeCQqW1lMqQbMkAw%3A1327016891064
|
|
Things working:
* you can create a report and assign runs to it
using the admin tool
* reports show up and you can view rough plots as well
as drill down to summaries of individual runs
Missing:
* flot needs support for:
- multibar grouping
- error bars
- angled text for labels
* the "Totals" table should be done as a percent
diff from the android toolchain
|