Tom Hoffman
2006-10-16 23:23:04 UTC
Hello all,
August has found SchoolTool in somewhat of a state of crisis. We are
(again) behind schedule, and while there are many small issues that
have caused delays, the overarching issue is that we planned and
budgeted for X hours of work from our various contractors, and we've
ended up with Y actual hours of work, where Y is about .5X. Beyond
that, I've got essentially no developers available now to get ready
for the start of the school year.
I'm not going to go into the gory details of how and why this
occurred, but there are a few implications:
* We are simply not in a position to follow through with comprehensive
testing at our partner schools at the beginning of the school year.
This is frustrating and embarrassing for us, and a disappointment for
everyone concerned. I did select schools that have existing systems
that they can continue to use in the fall, so it should be at most an
inconvenience and not a crisis.
* We have to conclude that the way we've organized this project isn't
working. Specifically, I'm not aware of another open source project,
especially one of SchoolTool's complexity, which has tried to use
teams of part-time paid contractors to write the code. We've been
able to write lots of good code, good components, but *finishing*
SchoolTool as a unified, coherent application has eluded us, and it
feels like without changes I could be writing exactly the same
apologetic email every six months.
So... Mark and Steve Alexander and I had a talk on Tuesday to plan out
some changes. These bits are still in the planning stages.
* We are going to hire a single full-time developer to handle the bulk
of SchoolTool development (instead of several part-time teams). I
would like this developer to also act more in the traditional role of
an open source project maintainer. I want to step away from trying to
be the mediator (or the "decider" as Bush would say) on technical
decisions. We may create a technical advisory board to help make
larger architectural decisions in the future.
* The full-time developer will work on site at a local school and will
create a complete version of SchoolTool for that school through the
next year. We need to eliminate the distance between our developers
and schools by getting them in direct physical contact.
* During the year we will use several development sprints to broaden
and generalize the application, and to create versions for the current
partner schools (if they will still have us).
I am also considering re-occupying my old office at my former high
school and spending much of my time implementing and customizing
SchoolTool there. I've been trapped in a state where I can generally
write code that works, but it is far to ugly and insane and generally
below the standards of SchoolTool development. I think it is time for
me to just pounding stuff out on my own development branch, whether it
ends up making it into SchoolTool proper or just acts as a prototype
for subsequent coding by real developers.
That's the working plan... still under development.
I want to stress that we have built a solid foundation for SchoolTool,
and attendance, calendaring and demographics are in pretty good shape.
We're not having huge technical problems. We just need to organize
the work differently and push this baby over the hump.
--Tom
August has found SchoolTool in somewhat of a state of crisis. We are
(again) behind schedule, and while there are many small issues that
have caused delays, the overarching issue is that we planned and
budgeted for X hours of work from our various contractors, and we've
ended up with Y actual hours of work, where Y is about .5X. Beyond
that, I've got essentially no developers available now to get ready
for the start of the school year.
I'm not going to go into the gory details of how and why this
occurred, but there are a few implications:
* We are simply not in a position to follow through with comprehensive
testing at our partner schools at the beginning of the school year.
This is frustrating and embarrassing for us, and a disappointment for
everyone concerned. I did select schools that have existing systems
that they can continue to use in the fall, so it should be at most an
inconvenience and not a crisis.
* We have to conclude that the way we've organized this project isn't
working. Specifically, I'm not aware of another open source project,
especially one of SchoolTool's complexity, which has tried to use
teams of part-time paid contractors to write the code. We've been
able to write lots of good code, good components, but *finishing*
SchoolTool as a unified, coherent application has eluded us, and it
feels like without changes I could be writing exactly the same
apologetic email every six months.
So... Mark and Steve Alexander and I had a talk on Tuesday to plan out
some changes. These bits are still in the planning stages.
* We are going to hire a single full-time developer to handle the bulk
of SchoolTool development (instead of several part-time teams). I
would like this developer to also act more in the traditional role of
an open source project maintainer. I want to step away from trying to
be the mediator (or the "decider" as Bush would say) on technical
decisions. We may create a technical advisory board to help make
larger architectural decisions in the future.
* The full-time developer will work on site at a local school and will
create a complete version of SchoolTool for that school through the
next year. We need to eliminate the distance between our developers
and schools by getting them in direct physical contact.
* During the year we will use several development sprints to broaden
and generalize the application, and to create versions for the current
partner schools (if they will still have us).
I am also considering re-occupying my old office at my former high
school and spending much of my time implementing and customizing
SchoolTool there. I've been trapped in a state where I can generally
write code that works, but it is far to ugly and insane and generally
below the standards of SchoolTool development. I think it is time for
me to just pounding stuff out on my own development branch, whether it
ends up making it into SchoolTool proper or just acts as a prototype
for subsequent coding by real developers.
That's the working plan... still under development.
I want to stress that we have built a solid foundation for SchoolTool,
and attendance, calendaring and demographics are in pretty good shape.
We're not having huge technical problems. We just need to organize
the work differently and push this baby over the hump.
--Tom