Mail Archives: geda-user/2016/07/11/00:54:02
On Mon, 11 Jul 2016, Evan Foss (evanfoss AT gmail DOT com) [via geda-user AT delorie DOT com] wrote:
> On Sat, Jul 9, 2016 at 8:10 AM, <gedau AT igor2 DOT repo DOT hu> wrote:
>> Hi all,
>>
>> there are more and more small, most often non-programming tasks in pcb-rnd
>> that could use your help (assuming you are a geda user).
>>
>> Please check out the list at:
>>
>> http://repo.hu/projects/pcb-rnd/help.html
>
> How many testers are you looking for?
"Never enough." Anyway I can try to give you estimations:
Number of systematic testers needed depends on the rate of development.
Made some stats recently, it turned out I spent more than 270 hours on
developng pcb-rnd this year, which results in a ~10 hours a week average.
One hour of development should be matched by one hour of testing at least.
It's totally unrealistic to find contributors with this much time at hand.
Assuming an average of 2 hours a week (an optimistic estimation), that
means 5 systematic testers, just to keep up with the current development
rate...
There's another very important segment, casual testers: they are
basically users, who use the software, preferrably in production. The
optimal number of them would depend on the size and compexity of the code
base, number of target platforms (which is more or less linux and modern
bsd, so 2), number of exotic features and strange use cases. I can only
guess, but for a pcb-rnd sized project I'd say at least 15..20 such
casal testers (production users) would be needed to roll out all
unexpected bugs.
So, a summary on what I'd find healthy, to include everything else:
1. a few dozen active occassional users; they try the software once, or do
a design once a year. They report only extreme breaks but sometimes also
report their ideas/needs.
2. 15..20 "power users", a.k.a. causal testers, a.k.a. production users.
They use the ware a lot and report anything that changed to the bad or
looks/feels strange. They have better understanding on the workflows and
thus they are in a very good position to cotribute valuable
ideas/suggestions
3. ~5 systematic testers; these are dedicated power users (surely
overlapping with group 2), who, beyond using the software, are willing to
spend some extra time on scheduled, systematic tests in order to increase
quality of the software
4. 1..3 developers; this is the number that'd be required to keep up with
bugreports and one new feature at a time, assuming the 2 hours a week
average. I'm doing this alone atm.
5. 1 or 2 non-developer crew, to maintain the homepage, documentation,
example/tutorial projects and whatnot - all the material that is essential
for the end user but is not in C. Again assuming 2 hours a week per
people. (And yes, I'm also doing this.)
5. 1..3 misc occassional but regular contributors; assuming way less than
2 hours a week, for things like "we need a new tool icon" or "there's this
open source EDA conference on the other end of the world, do we have
anyone there to attend and maybe give a speech about our stuff? And
others to make the presentation material? Maybe an event-specific page on
the web?"
6. 1 maintainers/coordinator/leader; mainly to give the whole thing a
direction and avoid... well, avoid what mainline demonstrates well:
team != a bunch or random people commiting in the same repository. Bert
took this role in mainline, tried hard for a long time, and I believe did
useful things. Lately I have the impression that he is giving up.
Anyway, this is all daydreaming today, pcb-rnd is almost a one-man show.
See my next mail.
- Raw text -