Do you refactor in small steps? - coding-style

Having read Fowler's "Refactoring" for a while, I still often catch myself thinking "I should have done this in smaller steps." -- even when I did not broke my code.
Refactoring in small steps is safe, but cost time. It's a trade off between speed and risk -- I try to be strategic in choosing the way how I am refactoring.
Nevertheless: Most the time I am doing refactorings in larger steps. If I took some of Fowler's "Mechanics" section and compare how I am working, I maybe find that I often leap two or five steps forward at once. This does not mean that I am a refactoring guru. My code maybe stay for 5 - 60 minutes broken or uncompilable.
Do you refactor in smaller steps and try to produce unbroken code in shorter frequencies? And: Are you successful in doing this?

Martin Fowler seems to lean towards the small, gradual refactoring approach. However, after reading his book he does occasionally make some drastic steps but only with unit tests to back up the code.
Refactoring is a controlled technique for improving the design of an existing code base. Its essence is applying a series of small behavior-preserving transformations, each of which "too small to be worth doing". However the cumulative effect of each of these transformations is quite significant. By doing them in small steps you reduce the risk of introducing errors. You also avoid having the system broken while you are carrying out the restructuring - which allows you to gradually refactor a system over an extended period of time. - Martin Fowler

I try :) The one urge I have to resist most while refactoring is actually making other changes along the way. Say I'm refactoring some code and see something unrelated in the code. I have to make a conscious effort not to go "fix" that as well. Make a note of it and move on. For one thing, it's a distraction from the task at hand. It also ends up polluting your change set so your commit message now has to document several seemingly random changes.

Yes, always. I think the real essence of refactoring is picking which steps to start with.
I find the thing with refactoring large changes in a safe manner is always to have a reasonably clear picture of where you want to go. Then consider your existing system and try to find out which pieces you can introduce that have least likelyhood of being a radical change. Then you can introduce these in a controlled and well tested manner.
So what you do is to work in the vincinity of the nastiness. Not always attacking directly from the front, but sometimes just chipping away small pieces. Usually I wait, and only go for the "big prize" after a few rounds of chipping away at minor nastiness. But I know where I want to go.
The nice thing about working this way is that you can maintain progress. You never "stop development to do refactoring". Arguably there are cases where stopping is the correct situation, but most of the time it's not.
The idea here is that if you "start" with cashing in the prize money, you will be spending the next X days doing the drudgery. And there's risk, maybe you chicken out or it doesn't work - or spend 6 months instead of a week. If you do the drudgery first, cashing in the prize will be possible with less risk. And your code will improve as you go. Sometimes you can decide that doing half the job was enough, since your understanding of the problem increases. Sometimes your idea of where you wanted to go was slightly botched, and you can realign your goal as you progress.
But its tempting to go straight for the reward.

I tend to refactor in large steps most of the time so I can see the forest from the trees. It's a "stream of consciousness" kind of programming. As long as you have your last working version safe in your source control of choice...

that's where the "red, green, refactor" approach is useful. At each stage you have the ability to verify that your code's behaviour is unchanged, and the refactoring only has to integrate the new behaviour.

The rule of thumb I use is refactor with tests and only refactor as much code as you are confident too.
At 60 minutes are you certain that your code is doing exactly what it should be. You'd need a lot of tests to pass. I would just try and get one going and then move on to the next.

If I have a clear picture of what I want to do, and if I can easily verify that I haven't broken anything afterwards, I am taking larger steps.
If the refactoring is more complicated, I try to break it down into smaller steps and do heavy testing after each one.

I usually refactor code as I change it. That is, instead of taking a piece of code and rewriting it while maintaining its function, I rewrite it towards a new functionality and in the process of doing so I improve the design of the code.
Often this means that by the time I've implemented the feature I was after I haven't done a complete and satisfactory refactoring of the old code. It is improved though, and I know I'll have the time to improve it further the next time I'm about to change its function.
For testing this means that I get to test both the refactoring and the new feature at the same time, which should save some time.
It also mean that I only spend enough time on refactoring to improve the maintenance situation required for that particular feature. This should help to avoid over engineering and/or wasting time refactoring stuff that already works and wont benefit from a better desing. By focusing only on code I would change anyway, there is also a high probability I will revisit that code in the near time to do further changes while it's in the users attention span.

Small discrete steps is what I'm most comfortable with, though at some points it can be a test of my self-control to reign in what could be a refactoring blood-bath. If I notice any improvements (no-matter how large) that could be made, I make a note of them and consider how it'd be split up into individual refactoring tasks. Plus, having a saga of changes in the commit message doesn't help.
NB. The code-base I work on is quite old, and full of those mystical bugs named after scientists. With large portions still lacking anything near even 50% test coverage it would be careless to get carried away.

Yep. I like to run the tests continually and so a chain of tiny refactors works well. I get really uncomfortable having my code broken for more than a few minutes at a time, and I generally revert if my code is broken when I go home at night, the re-write the next morning ALWAYS works better than trying to pick up where I was.

Related

When should you not refactor?

We all know that refactoring is good and I love it as much as the next guy, but do you have real cases where is better not to refactor ?
Something like time critical stuff or synchronization? Technical or human reasons are equally welcome. Real cases scenarios and experiences a plus.
Edit : from the answers thus far, it looks like the only reason not to refactor is money. My question is mostly relative to something like this: suppose you would like to perform "extract method", but if you add the additional function call, you will make the code slightly less faster and hinder a very strict synchronization. Just to give you an idea of what I mean.
Another reason I sometimes heard is that "others used to the current code layout will get annoyed by your changes". Of course, I doubt this is a good reason.
I'm a big fan of refactoring to keep code clean and maintainable. But you generally want to shy away from refactoring production modules that work fine and don't require change. However, when you do need to work on a module to fix bugs or introduce a new feature, some refactoring is usually worth it and won't cost much since you're already committed to doing a full set of tests and going through the release process. (Unit tests are very helpful, but are only part of the full test suite, as other posters noted.)
More significant refactorings may make it harder for others to find their way around the new code, and they may then react unfavorably to refactoring. To minimize this, bring other team members in on the process using an approach like pair programming.
Update (8/10): Another reason to not refactor is when you aren't approaching the existing code base with proper humility and respect. With these qualities you'll tend to be conservative and do only refactorings that really do make a difference. If you approach the code with too much arrogance, you may wind up just making changes instead of refactoring. Is that new method name really clearer, or did the old one have a name with a very specific meaning in your application domain? Did you really need to mechanically reformat that source file to your personal style, when the existing style met project guidelines? Again pair programming can help.
To reinforce the other answer (and touch on issues you mention): do not refactor a part of the code until it's well covered by all relevant kinds of testing. This doesn't mean "don't refactor it" -- the emphasis is on "add the necessary tests" (to do unit-tests properly may well require some refactoring, particularly the introduction of factory DPs and/or dependency injection DPs in code that's now solidly bolted to concrete dependencies).
Note that this does cover your second paragraph's issues: if a section of the code is time-critical it should be well covered by "load-tests" (which like the more usual kind, correctness-test, should cover both specific units [albeit performance-wise -- correctness-checking is other tests' business!-)] AND end-to-end operations -- the equivalent of unit tests and integration tests if one was talking about correctness rather than performance).
Multi-tasking code with subtle sync issues can be a nightmare as no test can really make you entirely confident about it -- no other refactoring (that might in any way affect any fragile sync that just appears to be working now) should be considered BEFORE one intended to make the synchronization much, MUCH more robust and sound (message-passing through guaranteed-threadsafe queues being BY FAR my favorite design pattern in this regard;-).
Hmmm - I disagree with the above (1st response). Given code with no tests, you may refactor it to to make it more testable.
You do not refactor code when you cannot test the resulting code in time to deliver it such that it is still valuable to the recipient.
You do not refactor code when your refactoring will not improve the quality of the code. Quality is not subjective, although at times, design may be.
You do not refactor code when there is no business justification for making an alteration.
There are probably more, but hopefully you get the idea...
As Martin Fowler writes, you shouldn't refactor if a deadline is near. That time in project is better suited to flush out bugs instead of improving design (refactoring). Do the refactoring omitted this time directly after the deadline is over.
Refactoring is not good in and of itself. Rather, its purpose is to improve code quality so that it can be maintained more cheaply and with less risk of adding defects. For actively developed code, the benefits of refactoring are worth the cost. For frozen code that there is no intention to do any further work on, refactoring yields no benefit.
Even for live code, refactoring has its own risks, which unit tests can minimize. It also has its own place in the development cycle, which is towards the front, where it's less disruptive. The best time and place for refactoring is just before you start to make major changes to some otherwise brittle code.
When it is not cost-effective. There's a guy at the place I work who loves refactoring. Making code perfect makes him very happy. He can check out a current project tree and go to town on it, moving functions and classes around and tightening things up so they look great, have better flow, and are more extensible in the future.
Unfortunately, it's not worth the money. If he spends a week refactoring some classes into more functional units that may be easier to work with in the future, that's a week's worth of salary lost to the company with no noticeable bottom-line improvements.
Code will never, ever be absolutely perfect. You learn to live with it, and keep your hands off something that could be done better, but perhaps isn't worth the time.
If the code seems very difficult to refactor without breaking, that's the most important code to refactor!
If there aren't any tests, write some as you refactor.
Honestly, the one case is where you are forbidden to touch some code by management/customer/SomeoneImportant, and when that happens I consider the project broken.
Here is my experience:
Don't refactor:
When you don't have test suite accompanying with the code you want to refactor. You might want to develop the test suit first instead.
When your manager doesn't really care about the maintainablity and extensibility of current code base, instead they care much about if they would be able to deliver the product on schedule, especially for the project with short and tight schedule.
If you stick to the principle that everything that you do should add value for the client/ business, then the times you should not refactor are the following:
Code that works and no new development is planned.
Code that is good enough / works and refactoring simply represents gold plating.
The cost of refactoring is higher than living with the existing code.
The cost of refactoring is higher that rewriting the code from scratch
Some of the other answsers say that you should not refactor code that does not have unit tests. If code needs refactoring, you should refactor it, you must however write tests first. If the code is written in a way that makes it difficult to test, it should be rewritten (in a perfect world).
When you've got other stuff to build. I always feel like refactoring an existing system when I'm supposed to be doing something else.
There's always a balance to be had between fixing or adding to code and refactoring. However, this balance is so far in favor of refactoring that I don't think I've ever been on a team that refactored too much. Chances are, if you think you're erring on the side of refactoring too much, you're right on the money.
Of course, the biggest determining factor is how close the deadline is. If a deadline is imminent, requirements come first.
Isn't the need to refactor code largely based on the propensity of people to cut and paste code rather than thinking the solution through, and doing the factoring in advance? In other words, whenever you feel the need to cut & paste some code, merely make that chunk of code a function, and document it.
I have had to maintain way too much code where people found it easier to cut and paste a whole function, only to make one or two trivial changes, which could easily have been parametrized. But like many other's experience, to try to refactor some of this code would have take a LOT of time and been very risky.
I have 4 projects wherein a 10K line collection of functions was merely copied and modified as needed. This is a horrid maintenance nightmare. Especially when the code has LOTS of problems, e.g. hard-wired endianness assumptions, tons of global variables, etc. I feel bile in my throat just thinking about it.
Don't refactor if you don't have the time to test the refactored code before release. Refactoring can introduce bugs. If you have well-tested and relatively bug-free code, why take the risk? Wait until the next development cycle.
If you're stuck maintaining an old flakey code base with no future beyond keeping it running until management can bite the bullet and do a rewrite then refactoring is a lose-lose situation. First the developer loses because refactoing bad flakey code is a nightmare and secondly the business loses because as the developer attempts to refactor the software breaks in unexpected and unforseen ways.
When you don't really know what the code is doing in the first place. And yes, I have seen people ignore that rule.
It's just a cost-benefit tradeoff. Estimate the cost to refactor, estimate the benefits, determine if you actually have the time to refactor given other tasks, determine if refactoring is the best time-benefit tradeoff. There may be other tasks more worth doing.

How to handle short coding sessions?

I'm working on some pet projects and generally I sit around my personal computer about 22:30 or 23:00 to code. But since I try to sleep about 24:00 I don't start coding and ending up reading articles, playing some games etc.
I don't feel like I can write decent code in an hour, because the project is quite big and I don't want to randomly or carelessly hack it. Even though I use TDD, most of the time stuff I'm doing is not straight forward which requires lots of testing before getting it right.
What's your approach to these kind of issues? Do you just code later when you got enough time or do you have a different approach which allows you to code just for 30 minutes and continue later?
I generally don't write lots of code until i have the time to do it. The reason is that for me to get effective takes focus and that takes a bit of time to be correctly focused. That said those 30min slots are great for
Writing more tests: nothing like trying to get to 100% code coverage, and it's not a big waste since you are investing
Research: I spend lots of time reading blogs, looking for frameworks I can use or tools. Spending 30min finding a framework that does 80% of a feature you need is much better than spending hours trying to code it. The other factor to this is that if you implement the framework and you find it is a bad fit you are better educated in the needs which means your development will be smoother.
Well my first thought was "use unit testing", but then I read you are already using this. But I still think it's the solution to your program.
Try to make your tests as small as possible and use the "1 assert per unit test" rule to create small atomic tests. You should be able to fix several of these small tests in a 30-minute session.
Here are some things you could try:
Don't sit near the computer. Instead, take a large piece of paper and go somewhere quiet. Think about what you want to accomplish. Write down interface ideas, detailed implementation. Make a list of questions you need to solve before you can go on.
Take off a week and code away. The ratio of getting-into-flow over flow-time is just too bad for 30 minutes.
Keep a log about what you do instead of coding. Observe your emotional state.
Go to bed early and try to have your pet coding session very early in the morning.
A small tip (that I use at work too) is to stop coding in the middle of something, with an obvious big red compile error waiting.
The next time you start working, the error will actually help you to remember what on earth you were doing.
While you are working on the small problem, the big picture clears up and then you can continue designing.
Developing with such short times is difficult, but you can still get something of that time. Unit testing is one. Writing down the interface of a class is another. While coding the real stuff will take much more time, these tasks are essentially a no-brainer, and they are just an exercise in typing.
So, my suggestion is: focus on small tasks that do not require thinking and concentration, and can be completed in the timespan you have.
Never worked for me, pet projects are usually too interesting and I end up working to late hours of the night or through a weekend.
I would suggest reconsidering your priorities - if all the time you have available is one hour late at night, maybe it's better to spend it on games, articles etc. Or just hanging out. When you have a bit more time, say a lazy Sunday, spend all the time at once and actually get the sense of accomplishing something that pet projects are supposed to give.
Here's what I do with my personal projects outside of work:
1) I try to give myself a good map of my project by planning it out on paper. I diagram all of my objects, data structures, and/or SQL tables and determine what basic functions and interactions between those components are necessary. I may write some actual code during this phase if the solution is obvious, but typically not.
2) Once the big picture is in place, I prioritize the most basic and critical elements. I also try to figure out which parts will be easier to write than others.
3) After priorities have been set I start working on the easiest and most critical parts first and work gradually towards the more complex and less important components. Breaking each task down into smaller parts tends to help. For example, I may design a database table first and the next day create a data interface class that controls interactions with that table.
4) Unit testing really helps me feel a sense of accomplishment, even if 30 minutes of effort only results in a few quality lines code.
5) Keep a change log, even if it isn't very detailed. I've found my change logs to be invaluable if I work on a big project in many short spurts over an extended period of time.
Those steps right there help me the most. In the end I am able to identify small chuncks of the project that can typically be completed in about 30-60 minutes. Of course, as a project develops, I usually have to re-evaluate something and go back to the beginning for a while when I discover I left something out of the planning phases. Sometimes I need to go a bit farther and give myself a time line with some deadlines and make sure I celebrate milestones with a personal reward. If you have a tendency of staying up till the wee hours of the morning, something I struggle with, I also suggest giving yourself a coding curfew as well. I also try to make sure my "coding computer" doesn't have many distractions on it, such as games.
There's always one more bug. If not, there's one more neat feature you can add, and THAT will add more bugs. Which is one reason why I think that the use of the phrase "All you have to do..." by anyone in (or using) IT should be a hanging offense.
I can cut down on how long my coding sessions last by doing trivial things, thinking things out while away from the keyboard (the shower is best - or while in bed at 4 a.m.) and by using lightweight environments such as script languages, but "quick" coding sessions are something I long ago gave up hope on.
Just shifting mental gears into the coding mode takes time, picking up the threads of where I was before takes time, discovering that my "quick and easy solution" was neither of the above takes time. Fixing my "quick and easy solution takes time, debugging - more time, and so forth.

How to refactor rapidly evolving code?

I have some research code that's a real rat's nest, with code duplication everywhere, and clearly needs to be refactored. However, the code base is evolving as I come up with new variations on the theme and fit them into the codebase. The reason I've put off refactoring so long is because I feel like the minute I spend a few days coming up with good abstractions, seeing what design patterns fit where, etc., I'll want to try out some new unforeseen idea that makes my abstractions completely inadequate. In other words, because of the rate at which the code is evolving, I really have no idea where abstraction lines belong, even though there is no shortage of (approximate) duplication and the general messiness of the code makes adding stuff to it a real pain. What are some general best practices for coping with this kind of situation?
Don't spend so long refactoring!
When you're about make a change in a piece of code, consider refactoring it to make the change easier.
After making the change, refactor again to clean up the damage done by that change.
In both cases, make the refactorings small and do them quickly, and move on.
You don't have to keep your code pristine at all times, but remember that it's easier to go fast if you have well-factored code to work in (and if you have good unit tests, of course).
Test Driven Development:
Red, Green, Refactor. Rinse, repeat.
Since it's one of the steps in every single cycle, you'll notice that's a LOT of usually minor refactoring taking place. That's the way it should be.
Your situation is pretty familiar to me. While doing investigative coding often you have no idea what the "right" abstraction will be, and as you say it can change with every new idea.Other posters have suggested:
Continuous small refactoring, which helps to avoid getting into the rats-nest situation
Test-Driven Development, which helps to find good, re-usable abstractions. It's important to note that TDD is less about testing than about doing good designs!
However, for investigative research code there is another strategy: the prototype. This seems to be what you are currently doing: coding as quickly as possible to prove a concept. There's nothing wrong with that, but a prototype should always be throw-away. Tweak it until you have all the necessary input and knowledge, then throw away the code and start over with TDD and continuous refactoring, and all your other "doing the things right" strategies.
Don't keep any of the code. Don't copy-paste anything. Don't refer back to it. Just start over with your new knowledge.
Clean up the code a little bit at a time. Always when you touch a class, try to leave the class cleaner that it was before you touched it ("the boy scout rule"). Refactoring is best done in very small steps, but very often.
Things like renaming some variable, splitting a method etc. take only some seconds or minutes. Large refactorings such as splitting or joining classes, may take an hour or two (and you make it in small steps, so that all tests pass at least every five minutes - otherwise you have entered Refactoring Hell and you should revert to the last known working state). If it takes days or weeks for you to refactor something, then it's not anymore "refactoring" - it's more like rewriting.
An article about this topic:
http://blog.objectmentor.com/articles/2007/07/20/whats-your-unit-of-measure
Put it in Distributed SCM like Git at least, that way when you break something refactoring you can reverse time divisibly to find the commit prior to the change, as well as being able to work on changes and commit them in branches without interfering with others work.
Gits Branch merge is great for things like this and you'll know easily if 2 people made incompatible changes in parallel without having to worry about the rest of the code.
For the above reasons, I would also create a seperate branch in the repository just for re factoring code with, and keep it up-dated regularly. This way, not only will others not interfere with your progress, but they can keep an eye on it and see changes in it that will eventually hit the main branch so they can pre-emptively code around those changes.
If you already know where there is duplication, you don't need several days to refactor it away.
Sometimes a rewrite is the only choice. This seems to be the case.
The CloneDR finds duplicate code, both exact copies and near-misses, across large source systems, parameterized by langauge syntax. It supports Java, C#, COBOL, C++, PHP and many other languages.
When it shows a parameterized abstraction of a set of found clones, it is essentially proposing that you refactor the code with that abstraction implemented (as a method, a function, a class, ...).
So running the CloneDR gets a list of potential abstractions to be added to your code, and replacing the clone instances by calls on the abstraction refactors your code thus cleaning it up (somewhat).
Even more remarkably, when it shows the parameter bindings used at each clone site needed to invoke the abstraction, it often shows a bungled clone instance, easily recognized when the bound paramters are conceptually inconsistent. If a parameer is bound to variables named YYYY-MM-DD, and one of them is YY-MM-DD, the "its a 4 digit-year" parameter type looks violated and in this this case there's a broken Y2K remediation. So examining the clone bindings often finds bugs.
This is a very common problem in scientific computing. Some of the most effective ideas for reducing the size and complexity of code require leveraging assumptions, and science demands that you constantly change those assumptions.
All you can do is try to refactor your code as you go, and try not to write yourself into any corners. Also work with good people who understand the value of not making a mess.

At what point does refactoring become not worth it?

Say you have a program that currently functions the way it is supposed to. The application has very poor code behind it, eats up a lot of memory, is unscalable and would take major rewriting to implement any changes in functionality.
At what point does refactoring become less logical then a total rebuild?
Joel wrote a nice essay about this very topic:
Things You Should Never Do, Part 1
The key lesson I got from this is that although the old code is horrible, hurts your eyes and your aesthetic sense, there's a pretty good chance that a lot of that code is patching undocumented errors and problems. Ie., it has a lot of domain knowledge embedded in it and it will be difficult or impossible for you to replicate it. You'll constantly be hitting against bugs-of-omission.
A book I found immensely useful is Working Effectively With Legacy Code by Michael C. Feathers. It offers strategies and methods for approaching even truly ugly legacy code.
One benefit of refactoring over rebuilding is that IF you can do refactoring step by step, i.e. in increments, you can test the increments in the context of the whole system, making development and debugging faster.
Old and deployed code, even when ugly and slow, has the benefit of having been tested thoroughly, and this benefit is lost if you start from scratch.
An incremental refactoring approach also has helps to ensure that there is always a product available which can be shipped (and it's improving constantly).
There is a nice article on the web about how Netscape 6 was written from scratch and it was business-wise a bad idea.
Robert L. Glass suggests that
Modification of reused code is particularly error-prone. If more than 20 to 25 percent of a component is to be revised, it is more efficient and effective to write it from scratch.
Well, the simplest answer is if it will take longer to refactor than it will to rebuild, then you should just rebuild.
If it's a personal project then you might want to rebuild it anyway as you will probably learn more from building from scratch than you would from refactoring, and that's one big objective of personal projects.
However, in a professional time-limited environment, you should always go with whatever costs the company the least amount of money (for the same payoff) in the long run, which means choosing whichever takes less time.
Of course, it can be a little more complicated than that. If other people can be working on features while the refactoring is being done, then that might be a better choice over having everyone wait for a completely new version to be built. In that case rebuilding might take less time than just the refactoring would have taken, but you need to take the entire project and all contributors of the project in to account.
When you spend more time refactoring than actually writing code.
At the point where the software doesn't do what it's supposed to do. Refactoring (changing the code without changing the functionality) makes sense if and only if the functionality is "as intended".
If you can afford the time to completely rebuild the app, don't need to improve functionality incrementally, and don't wish to retain any of the existing code then rewriting is certainly a viable alternative. You can, on the other hand, use refactoring to do an incremental rewrite by slowly replacing the existing functions with equivalent functions that are better written and more efficient.
If the application is very small, then you can rewrite it from scratch. If the application is big, never do it. Rewrite it progressively, one step at a time validating you didn't break anything.
The application is the specification. If your rewrite it from scratch you will most likely run into a lots of insidious bugs because "no one knew that the call to this function was supposed to return 3 in that very specific case" (undocumented behaviour...).
It's always more fun to rewrite from scratch so your brain might trick you into thinking it's the right choice. Be careful, it's most likely not.
I've worked with such applications in the past. The best approach I've found is a gradual one: When you are working on the code, find things that are done multiple times, group them together in functions. Keep a notebook (you know, a real one, with paper, and a pencil or pen) so that you can mark your progress. Use that in combination with your VCS, not instead of it. The notebook can be used to provide an overview of the new functions you've created as part of the refactoring, and the VCS of course fills in the blanks for the details.
Over time, you will have consolidated a lot of code into more appropriate places. Code duplication during this period of time is going to be next to impossible, so just do it as best as you can until you've reached a point where you can really start the refactoring process, auditing the entire code base and working on it as a whole.
If you've not enough time for that process (which will take a very long time), then rewriting from scratch using a test-first approach is probably better.
One option would be to write unit tests to cover the existing application and then start to refactor it bit by bit, using the unit tests to make sure everything works as before.
In an ideal world you'd already have unit tests for the program, but given your comments about the quality of the app I'm guessing you don't...
No document, no original writer, no test case, and a bunch of remaining bugs.
Uncle Bob weighs in with the following:
When is a redesign the right strategy?
I’m glad you asked that question. Here’s the answer. Never.
Look, you made the mess, now clean it up.
I’ve not had much luck with small incremental changes when the code I inherit is really bad. In theory the small incremental approach sounds good, but in practice all it ends up with is a better, but still poorly designed application that everyone thinks is now YOUR design. When things break, people no longer think it is because of the previous code, it now becomes YOUR fault. So, I would not use the word redesign, refactor or anything else that implies to a manager type that you are changing things to your way unless I was really going to do it my way. Otherwise, even though you may have fixed dozens of problems, any problems that still existed (but weren’t discovered) are now going to be attributed to your rework. And be assured that if the code is bad then your fixes will uncover a lot more bugs that were simply ignored before because the code was so bad to begin with.
If you truly know how to develop software systems then I would do a redesign of the whole system. If you don’t TRULY know how to design GOOD software then I’d say stick with the small incremental changes as you may otherwise end up with a code base that is just as bad as the original.
One mistake that is often made when redesigning is that people ignore the original code base. However, redesign does not have to mean totally ignore the old code. The old code still had to do what your new code has to do, so in many cases the steps you need are already in the old code. Copy and Paste then tweak works wonders when redesigning systems. I have found that in many cases, redesigning and rewriting an application and stealing snippets from the original code is far quicker and much more reliable than small incremental changes.

Programming in a crunch

Ideally you want a schedule that's accommodating and flexible but when it comes to paying the bills and working in a business, that's rarely a luxury programmers have.
I have been fortunate to have the grace of Steve McConnell and Frederick Brooks to tell me what to do if I want to screw up my project and I take their work seriously.
And yet there are still times when your back is against the wall and you need to speed up the work. What are some "tweaks" to your process you've used to speed up delivery without sacrificing quality? Is this even possible?
I consider learning good practices to be of higher value than learning the actual code so no "be a better coder" answers. That's a given.
jpgs are faster than html pages
Without sacrificing quality, your best options are pretty much the same best options that you have when it's not crunch time: Eliminate unnecessary distractions, focus on the most important work first, obtain better hardware, and offload any work that you can reasonably offload.
Arrange my tasks in some sort of list of small independent tasks (or, tasks where the dependencies are already taken care of in the order). Then, just sit down and clear one task off my plate after another. Anything where I have to coordinate with someone else, ask a question, etc. is punted forward; I just sit and code with no distractions, until I get to a place where I need some sort of outside interaction. Then I deal with all of my coordination with other people, re-planning, and so on in one batch, and go heads down into straight coding again.
Also, prioritize and cut everything that's optional. If you get the required stuff done in time, then go back to the optional stuff, but do what is absolutely required first. This may be relaxed slightly if there are tasks that are easier to do when you're in the right context, but on the whole, try to keep the optional parts to a minimum.
Oh, and don't make these judgement calls while you're coding. When you're coding, just take one task off your list, do it, and move on to the next. As I said, batch up all of the non-coding work into chunks so it doesn't slow your coding down.
Anyhow, that's what I do when I'm up against a wall. Not sure how well it works for anyone else.
Usually, cutting features and work items is the best way to meet a tight deadline.
You can try to rush coding, but you'll almost always pay for it later - You'll spend much more time debugging and stabilizing the rushed code than you saved in the first place.
Actually, I think a lot of programmers and organizations have a lot of low-hanging fruit. Make sure you are using your time productively, know how to trade-off speed for quality appropriately, don't "gold plate" things beyond what your customer is asking for, etc.
Profile your organization's processes. What is the bottleneck slowing you down? Context-switching (i.e., multitasking) is often a real productivity-killer.
I suppose the problem with this is that when your back is up against the wall, organizational improvements like these seem like luxuries unto themselves, even though they are the things that could really help you speed up your work.
The biggest help (for me) is forcing myself to complete each single task before moving on to another one. If I need to get some bugs out of libfoo, or even write libfoo to go on to the next step, I force myself to stay on that task, then the next one, then the next one.
Most of the time that I find myself in a crunch, its because I was too much of a grasshopper, jumping around from thing to thing.
This morning I sat down and forced myself to write a bunch of unit tests, rather than working any more on the library being tested. It sucked, but now they're done, and I can really sail through finishing my library.
I'd suggest learning GTD or something like it before you get in the crunch, using to avoid a crunch if at all possible and then, if you must, get through the crunch by just doing it. Get in the zone and persevere.

Resources