You’ve finished the design for an Apex application, and the manager asks you “when will you have it ready to test”. You resist the temptation to respond snarkily “how long is a piece of string” – which, by the way, is often the only appropriate answer if they ask for an estimate before the design work has started.
Since you have a design and a clear idea of what exactly this application will do, you can build a reasonable estimate of development time. The starting point is to break down the design into small chunks of discrete modules of work, where each chunk is something you can realistically estimate to take between half a day up to a maximum of three days. In cases where it’s something you haven’t done before, you can reduce uncertainty by creating a small Proof-of-Concept application or code snippet and seeing how the abstract ideas in the design might work in reality. This is where Apex comes in handy – by the time the design has completed, you’ll have a database schema (you created the schema in a tool like SQL Developer, didn’t you – so you can generate the DDL for that in a jiffy) and you can just point Apex to that and use the New Application wizard to create a simple application.
I plan all my projects using Trello, recording estimates using the Scrum for Trello extension. I don’t know much about the “Scrum” method but this extension allows me to assign “points” to each task (blue numbers on each card, with a total at the top of each list). I used to use a 1 point = 1 day (8 hours) convention, but I was finding most of the tasks were more fine grained (I was assigning 0.5 and 0.25 points to most tasks) so I’ve now switched to a convention of 1 point = 1 hour (more or less). In other words, I’d report my estimates with the assumption that my “velocity” would be 8 points per day.
My points system looks roughly like this:
Simple report, form or process = 4 points
Complex report or form or process = 8 points
Very complex form or process = 24 points
Simple bug fix / tweak = 1 point
Complex bug fix / enhancement = 2 to 8 points depending on complexity
In addition to that, I keep a daily diary of work done in Evernote which has allowed me to do what I’ve wanted to do for a while now: measure my actuals vs. estimates for my projects, in order to calibrate my estimates for future projects. I made up a quick spreadsheet showing the development work for two projects, showing the original estimate, the start and finish dates, and actual development days worked (accurate roughly to the nearest half day, not counting non-development time like meetings and other project work). This allows me to see how my actual development time compares to my original estimates.
Since my reported estimates were roughly 8 points = 1 day, it seems I tend to underestimate. It may seem that my estimates are getting better (since my Points per Day is approaching 8), but that trend is more likely a result of SUM involving a lot less uncertainty and risk and being a smaller project overall. SAM was in a new environment and new client, whereas SUM is merely an additional application in the same environment for the same client. I was also able to re-use or adapt a lot of infrastructure-type code and Apex components from SAM in SUM.
The other thing that I can see from the details is that my “velocity” (points per day) is higher in the earlier stages of development, probably because I have plenty of work planned out, and I can work for several days in a row, uninterrupted by meetings and other distractions. In later stages, my attention gets divided by bug fixes, enhancement requests, meetings, doing design work for the next project, and general waiting-for-other-people-to-do-their-jobs.
For my next project I’ll estimate using the same points system as above, but factor in a 7 points-per-day to the estimates that I report.
Do you have a system of your own you use to create estimates and measure their accuracy? If not, why not try this for yourself? Do you keep track of your estimates and progress? If not, I recommend you start :)
I have Tom Kyte’s “Ask Tom” in my feed reader, and every now and then it opens for questions and the flood pours in. Quite often there is an interesting question, but more often than not there are quite a lot of “poor” questions: some just poorly worded, some too-easily-googleable, others are could-have-answered-from-the-docs questions. It’s not dissimilar to StackOverflow, which doesn’t suffer so much because of the army of people who work together to either improve or close these types of questions. Sometimes I think “Why did you waste Tom’s time with that? You could have googled it, searched the docs, or opened SO and you’d probably find a good answer there without even having to ask it.”
However, my impression is wrong, I think, because my feed reader only shows me the initial question, and not the detailed follow-up by Tom and others. I was reminded today of how a poor question can still lead to enlightenment, by this excellent quote by Tom:
As things grow and change over time – engineers need to use different approaches. The planes I fly in have wings that go straight out from the plane. That is because I fly at about 500 mph. A fighter jet at mach 2 would be going 1,500 mph. The wings on a fighter jet are significantly different in their architecture – because if they weren’t it wouldn’t work. We have different types of bridges in order to cross different types of chasms. We have different building styles for different types of builds (stick and brick for a 2 story house, steel girders in a frame for 30 story buildings).
As you scale up, the solutions that worked for trivial amounts of data will not necessarily work for large amounts of data.
If you are in a place that says “make it work but do not change code” – expect to have a bad day.
It’s that wonderful time of year again – no, not talking about springtime, when I’m stalked by airborne grass pollens causing me to choke and gag all day. The sound of a lawnmower is enough to elicit a sneezing fit – and in Aussie springtime, that’s most weekends. (I’m so glad my wife does the mowing at our place, usually while I’m at work. My son is now old enough to do the mowing – and I’m quite happy to pay him to do it, I can tell you!)
The “wonderful time of year” I referred to earlier is that time when the excitement of OpenWorld has petered a bit and AUSOUG have released the program for Perth’s excellent conference.
I’ll be presenting a talk about Building Maintainable Applications in Apex, which I sincerely hope will be a lot more interesting than the title sounds. (It’s funny how you think after rejecting several competing wordings, the title you finally settle on is perfect when you submit the abstract – but when you see it alongside the other gems on the program you think “surely I could have come up with a more interesting title?”. Oh well.)
I’ll be talking about why you should remove most of the business logic from your Apex application – or at least consider it. I won’t promise it will necessarily be earth-shattering or original. I think there’s some ideas here that are worth spreading around – get people thinking, talking and writing about. Most of the talk will be in reference to a recent greenfields project, but the ideas are ones I’ve picked up over the years at different locations and from other Oracle developers – bits that I find work well or seem worthwhile expanding on.
If that topic doesn’t grab you, you might be interested in hearing Tim Hall talk about virtualization. If I wasn’t speaking at the time I would have attended that one – so if you do, could you make sure to take a few notes for me? Thanks.
I hope to see you there. But – fake flowers only please.
If you’re interested in my presentation on the Function Result Cache, it’s now available from my presentations page. It was given this morning at Oracle’s offices in Perth to the local AUSOUG branch and seemed to go down well and I got some good feedback. It was only a little overshadowed by all the hoopla over the release of 12c :)