|
- If your team schedules 160 points, but can only deliver 40, the first thing you need to fix is to stop scheduling 160 points. 160 sounds suspiciously like 4 people x 40 hours a week. In case you haven't heard, people don't spend 8.000 hours a day coding. The go on vacation. They go to meetings. They get pulled off coding to fix tooling issues. They get pulled off to fix bugs. They yak about the Superbowl. But no matter what your points are, your record of delivery is the best estimate of your future capability.
- If you have a critical delivery in a couple of short months (counting the Christmas holiday no less), it's time to 'fess up to senior management and your customers right now so they can mitigate the negative effects of the code being late. Abandoning your process right before a big deliverable is not a highway to success. Changing your process will make your schedule more uncertain, not less.
- Writing tests may take as much time as developing the code; that's just the way it is. Test-first is a great way to wring out a design, too. Tests are not what the Agilists call "documentation". They are talking about 500 page Software Requirement Specifications and such. You complain that bugs are escaping module test and getting into demos. Abandoning your testing process is unlikely to make that better.
Longer term, there may be some positive changes you can suggest to your team. Trust the team. If they want to do the right thing, they will want to improve on weaknesses in their process.
- Make the team schedule larger refactorings as if they were story points. This will increase the visibility of what they are doing. If all they want to accomplish is refactoring, then maybe there's something wrong upstream, like poor design. Test-first is supposed to help wring out the design. If it's not working in this team, ask the team if it is valuable enough to retain.
- Consider adding code reviews. Don't allow refactorings to be checked in if the team doesn't agree that it improved something. People hate working for a week and having it thrown away. Once this happens a few time, your team will become more focused on the need for any change they propose. Consider design reviews too, to make sure code needs refactoring before refactoring it.
- Engage the team in a conversation about the bugs that escape module test. Honestly, my experience is that good module test finds most new bugs. Maybe some modules are not well tested. Maybe some modules are a big sticky mess that is difficult to test. If so, maybe that is what needs refactoring. If that's not feasible, maybe it's time to invest in some dedicated testers to do system test on a recurring basis. If your bugs are in a tracker of some sort, it's possible to do a causal analysis to see what part of your process injects the most bugs. Maybe introducing static checkers...
Most importantly, management must lose their ego and stop blaming the team. Software takes as long as it takes. If it's taking 4x as long as you want it to take, this is more likely a management problem than a team problem. It means you need 4x (or 5x) the team. It means spending on better testing. It means investing in more tracking, more analysis, and more process. Or it means scaling back your expectations.
|
|
|
|
|
SeattleC++ wrote: 60 sounds suspiciously like 4 people x 40 hours a week
We have about 15 devs (me included).
SeattleC++ wrote: Writing tests may take as much time as developing the code; that's just the way it is.
It's taking about 10x the time it takes to write the code.
SeattleC++ wrote: Make the team schedule larger refactorings as if they were story points. This will increase the visibility of what they are doing. If all they want to accomplish is refactoring, then maybe there's something wrong upstream, like poor design. Test-first is supposed to help wring out the design. If it's not working in this team, ask the team if it is valuable enough to retain.
Yes, I think this is happening a lot..
Ah, I see you have the machine that goes ping. This is my favorite. You see we lease it back from the company we sold it to and that way it comes under the monthly current budget and not the capital account.
modified 31-Aug-21 21:01pm.
|
|
|
|
|
It is possible to have too many unit tests. Especially if they are testing lots of internal functions. Mostly you want to just test the external interfaces and if the internal functions are working properly the external ones will too.
And of course as you've found out automated tests are great but can never replace an actual person using the product. You need testers, even if it's the CEO and receptionist.
Agile is not against design, btw. Spend some time locking down those external interfaces and then DON'T CHANGE THEM. Combined with unit tests only on those external things, you can develop really quickly by relying on the mostly static tests to catch regressions.
Stick to it. Even if you find out your external interface design could be better, unless it's really screwed up STICK with it. Fix everything in version 2.
And exercise more control during story creation to keep people on track. If something works already, keep it and work on something that is broken.
|
|
|
|
|
Agile is evil. It no longer values the coders and ends up in a blather fest. Let the non technical types have meetings and then put someone with tech knowledge in charge of assigning the work and coordinating the project. He should also have coding work. Assign someone you don't like to speak with the blatherskites.
Agile breeds black holes.
|
|
|
|
|
It sounds like they aren't breaking problems up appropriately. It also sounds like the team is too large. We have a team size of only 2 developers, but we never take anything over 8 points and we are very cautious about those 8. We also rarely take over about 20 total points per sprint. If they only deliver 40 per sprint don't let them commit to more than that. Something else can always be pulled in later in the sprint if they get done early.
Also, "Needs to be delivered by XX date" is almost always a bad way of thinking. It will cause a lot more defects and a lot more problems, and will usually cause the software to be delivered LATER than it could have been. Software is done when it's done. You may want to go back and say, "Well, we can't have the entire project by that date, but what is the minimum that will still be usable?" They will almost always say, "All of it," and then you get to say, "Well, then it will be YY date, at the very earliest, but we can give you (some subset of user stories) by XX date and deliver the rest later."
|
|
|
|
|
What I've seen is programmers now who are terrified of writing code. They don't feel like they can write a simple for loop without a unit test (written first!) to show it works.
Then you take all the unit-test passing code and plug it together and find it is all still buggy because unit tests don't catch everything and the most subtle bugs occur in integration (especially with anything event driven).
My favorite sayings are "you don't ship tests" and "customers don't buy tests". What ultimately matters to the business is to ship working code. Absolutely write tests for subtle algorithms or pieces of code that have proven buggy. Writing tests for code that has to be refactored (for performance for instance) is great to prove that the behavior is unchanged. But it seems the modern programming culture puts all the emphasis on unit testing as if that is a panacea for delivering working systems.
And don't forget that writing test suites for code that doesn't ship due to being un-needed or because it is delivered too late is completely wasted time.
Maybe these guys haven't been on a project that failed due to being delivered late, and they have to learn the hard way.
|
|
|
|
|
Great comments, I agree 100%..
Carlosian wrote: Maybe these guys haven't been on a project that failed due to being delivered late, and they have to learn the hard way.
Their previous project took 4 years to fail and cost around £25 million. Odd how they haven't learned anything from that! Even odder that they weren't all kicked out.. 
Ah, I see you have the machine that goes ping. This is my favorite. You see we lease it back from the company we sold it to and that way it comes under the monthly current budget and not the capital account.
modified 31-Aug-21 21:01pm.
|
|
|
|
|
The bad news: You are probably toast.
The good news: You know this now, not on the ship date.
Agile is not a magic wand, you still need engineering skill.
However, Agile processes are designed to expose problems. It worked.
Given lots of test effort and still shipping bugs it sounds like a competence problem.
How to deal with that is up to you and your organization.
Sorry I can't be more specific as I have not hit that particular problem.
|
|
|
|
|
Maybe you've got a few things the wrong way around.
Are you asking too much of the team in too short a space of time?
If this sprint and the one prior is only delivering 40 points instead of the anticipated 160, then perhaps that is too much. Even if you didn't do unit testing and just went straight to system testing would the 160 be realistic?
Remember to apply S.M.A.R.T. objectives for your team. That might help keep things focused. You're probably doing this anyway, but direct your team to complete the most critical changes in a sprint first, so the least critical changes can roll over. If you're delivering the important stuff then perhaps your management will overlook the shortfalls in delivery. At least for a while.
Take another look at the project plan/road-map. Has enough time been allocated to testing? I'd say a rule of thumb would be 1 hour of dev should have 1 hour of testing. Even better to have 2 hours of testing. You could consider the act of maintaining unit tests goes into part of the testing time. Maybe that is the difference between 160 and 40 points - there is a whole bunch of time for unit test development that isn't accounted for. You may be able to justify an extension to the development based on actual performance and gaps in the project plan.
If the subject of extending the team comes up in that context, definitely go for more testers, or more developers dedicated to supporting the testers and/or unit tests. Testers and junior devs tend to be cheaper so you could get more bang for your buck there.
As you've pointed out in your post, a lot of time can be sunk into testing one way or another. Even if you didn't use unit testing & CI there would still be a test team that is likely to not inherently have the skills necessary to reset systems, write scripts and generate test data. Inadvertently the testers could be a drag on your dev team, so make sure they have access to dev skills. Even if that means assigning a developer specifically to that role. To put it another way, sacrifice one dev to testing so everyone else can get on with their jobs!
If this still doesn't get the delivery you want, then don't be afraid to ditch the unit testing entirely. Your team will hate it. Maybe it will bite you in the arse later; but later is for those that survive the deadline. However, if you do, then make double sure that your system testing is giving you good coverage.
Finally, my old PM used to say that the first thing he did with a failing team was to reduce (not increase) headcount. I think there's been some studies on this, but basically those that are left behind tend to have tighter communication channels with each other, where they don't spend time having large meetings or email discussions. For whatever reasons, it's an opportunity to trim out team members that aren't a good fit for the team in that particular project environment (high-pressure & short lead times). I think it was about 5 devs to be optimal, but be careful to do some more reading on this subject.
That's it. All in my opinion. Take it or leave it as you see fit. And Good Luck!
|
|
|
|
|
So, we have a Secret Santa thing going on at work through Elfster[^].
This is my wish list: [^]
For some reason the file has gone missing: here is the list:
REI gift card
Time machine - prefer TARDIS
Phased plasma cannon in the 40 watt range
Working Starship with warp drive
I can't imagine I'll get my first pick... 
What did you ask for?
modified 9-Dec-16 13:51pm.
|
|
|
|
|
I can't imagine you'll get any of those unless Santa is real... please, please let him be real!
If so, can I get a ride in your Tardis?
- I would love to change the world, but they won’t give me the source code.
|
|
|
|
|
Forogar wrote: If so, can I get a ride in your Tardis?
I told you next Tuesday that I'd pick you up last Thursday and you weren't there!
|
|
|
|
|
Few (if any) of the items on my wish list fall in under the KSS category.
| Ravings en masse^ |
|---|
| "The difference between genius and stupidity is that genius has its limits." - Albert Einstein | | "If you are searching for perfection in others, then you seek disappointment. If you are seek perfection in yourself, then you will find failure." - Balboos HaGadol Mar 2010 |
|
|
|
|
|
|
This www.codeproject.com page can’t be found
No webpage was found for the web address: https://www.codeproject.com/script/Membership/Uploads/46383/Untitled.png
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
|
|
|
|
|
Odd. Seems to be a bit flaky! Wait, it's gone! How bizarre!
Here is what I asked for:
REI gift card
Time machine - prefer TARDIS
Phased plasma cannon in the 40 watt range
Working Starship with warp drive
|
|
|
|
|
That's a pretty weedy plasma cannon - you sure it wasn't 40 Gigawatt you wanted?
40 watt is the JML / Lidl / Walmart Black Friday of plasma cannons.
And isn't the Tardis usually full of hidden daleks?
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
|
|
|
|
|
OriginalGriff wrote: That's a pretty weedy plasma cannon
If it was good enough for the Terminator...
|
|
|
|
|
Never go clubbing in a new car: the warrantee is void if the seal is broken.
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
|
|
|
|
|
That would not get my squeal of approval.
The language is JavaScript. that of Mordor, which I will not utter here
This is Javascript. If you put big wheels and a racing stripe on a golf cart, it's still a f***ing golf cart.
"I don't know, extraterrestrial?"
"You mean like from space?"
"No, from Canada."
If software development were a circus, we would all be the clowns.
|
|
|
|
|
PETA called. They won't this TOTD retracted. It doesn't meet their seal of approval.
|
|
|
|
|
Squeal of approval, if anything.
The language is JavaScript. that of Mordor, which I will not utter here
This is Javascript. If you put big wheels and a racing stripe on a golf cart, it's still a f***ing golf cart.
"I don't know, extraterrestrial?"
"You mean like from space?"
"No, from Canada."
If software development were a circus, we would all be the clowns.
|
|
|
|
|
Folks harp about that all the time.
"the debugger doesn't tell me anything because this code compiles just fine" - random QA comment
"Facebook is where you tell lies to your friends. Twitter is where you tell the truth to strangers." - chriselst
"I don't drink any more... then again, I don't drink any less." - Mike Mullikins uncle
|
|
|
|
|
|
If you keep both eyes open; looks like a goat. Pink goat. 
The sh*t I complain about
It's like there ain't a cloud in the sky and it's raining out - Eminem
~! Firewall !~
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
