Hi, Indeed, this was a question I asked myself as well. And I think I figured out a way to filter most of the false positives out. The analysis goes like this: If you have a misstep, you likely want to restart Spotter and search again. In this situation it is likely that the distance between the end of the misstepped session and the starting of a new session is less than 5 seconds. For safety, letâs consider 20 seconds. To put it in perspective, 90% of Spotter sessions have less than 20 seconds. So, if we filter out all sessions that have another session coming after less than 20 seconds we get this: 'Classes'->40 'Implementors'->38 'History'->33 'Packages'->20 'Menu'->18 'Messages'->18 'Catalog Projects'->12 'Instance methods'->10 'Senders'->7 'References'->5 'Playground named pages'->5 'Pragmas'->4 'Playground cached pages'->3 'Selectors'->3 'Super instance methods'->3 'Help topics'->3 'ws.stfx.eu'->2 'Examples'->2 'GitHub Baselines'->2 'Class methods'->2 'Global variables'->1 'All subclasses'->1 'Files'->1 'Monticello Repositories'->1 'Metacello Configurations'->1 'Class instance variables'->1 'Items'->1 'Help contents'->1 'Productions'->1 'Dirty Monticello packages'->1 There are very little changes. History is still number 3, and Menu is in top 5. Cheers, Doru
On Jun 10, 2016, at 8:42 AM, Max Leske <maxleske@gmail.com> wrote:
Hi Doru,
I just want to point out that history may possibly be a bit of a false positive. When I open spotter, type and hit enter quickly I sometimes hit an entry from history that I didnât intend to (one annoying example of this is ProcessBrowser which I regularly hit accidentally when searching for Process).
Cheers, Max
On 10 Jun 2016, at 00:56, Tudor Girba <tudor@tudorgirba.com> wrote:
Hi,
Juraj, Andrei and I did a rough analysis collected from 94 computers over the past 7 months. Of these, only 42 recorded more than 9 sessions so we only focused on these. It can be because the rest switched off the data collection in the meantime. We also excluded the computers of the GT team members.
Of these 34 used the dive-in feature. That is, these users used at least one contextual search.
We looked at the event of acting on an element (pressing Enter), and we collected the parent category. Acting on an item indicates that intent of search. There are 35 categories used in total, with 8 being used by 10 people (25% of the studied population) or more. Below you can see also the amount of computers using it:
<spotter-categories-distribution.png>
'Classes'->40 'Implementors'->38 'History'->34 'Menu'->24 'Packages'->23 'Messages'->19 'Catalog Projects'->12 'Instance methods'->10 'Senders'->9 'Pragmas'->6 'References'->5 'Playground named pages'->5 'Playground cached pages'->4 'Help topics'->4 'Examples'->3 'Super instance methods'->3 'Selectors'->3 'ws.stfx.eu'->2 'GitHub Baselines'->2 'Dirty Monticello packages'->2 'Class methods'->2 'Global variables'->1 'All subclasses'->1 'Example Subjects'->1 'Files'->1 'Monticello Repositories'->1 'Metacello Configurations'->1 'Class instance variables'->1 'Items'->1 'Tags'->1 'Help contents'->1 'Monticello Package'->1 'Instance variables'->1 'Productions'->1 'Methods'->1
Also, in this analysis, some of the categories appear also at deeper levels (Senders, Implementors, References, Instance methods).
As expected, Classes and Implementors are on top. Yet, the third is History, and it is also interesting to see that there is a high usage of a search through the World menu elements, but also of the Packages.
We also note that there is quite a long tail, and this seems to confirm the hypothesis that different people have different needs and that these differences should be supported by the IDE.
This analysis was carried out using the code that Juraj and Andrei put together for analyzing the data from the event recorder.
Cheers, Doru
-- www.tudorgirba.com www.feenk.com
"Value is always contextual."
-- www.tudorgirba.com www.feenk.com "Innovation comes in the least expected form. That is, if it is expected, it already happened."