In a critical part of my project which basically allows objects to be received by a controller asynchronously, put into a Queue, processed sequentially from the queue one at a time by a thread, then service responds, older processed objects are kept in the queue until newer item insertion.
Back in time (months ago), my Queue implementation for solving this particular business specific issue behind this was to use Guava's EvictingQueue, which now is marked as #Beta, and so this part of the application can break in future Guava releases.
private final Queue<SomeRandomBusinessObject> items = Queues.synchronizedQueue(EvictingQueue.create(queueSize));
Are there any thread-safe and fixed-size alternatives to EvictingQueue to achieve this goal?
There are couple of inaccuracies / mistakes in your post, so let's just try to find common ground.
First, any new feature in Guava is annotated as #Beta from the beginning, same is true for EvictingQueue in 15.0 (this links to 15.0 docs). So you probably missed that fact couple months ago, but that's OK, because...
...#Beta doesn't really mean it'll be changed without any notice -- on the contrary, some time ago, after some feedback from the community, Guava devs established pretty strict policy about what and when can be changed. See PhilosophyExplained wiki page, which says (emphasis mine):
Beta APIs
Beta APIs represent Guava features that we aren't ready to freeze for whatever reason: because the methods might not find enough users, because they might be moved, because their uses might be too narrow to include them in Guava.
That said, #Beta APIs are fully tested and supported, and treated with all the care and affection that the rest of Guava receives.
This means EvictingQueue quality is not worse than if it wasn't a "beta feature".
The biggest connotation of the #Beta annotation is that annotated classes or methods are subject to change. They can be modified in any way, or even removed, at any time. If your code is a library itself (i.e. it is used on the CLASSPATH of users outside your own control), you should not use beta APIs, unless you repackage them (e.g. using ProGuard).
This could be the concern you brought up when talking about "braking up in the future", but...
All this said, #Beta features tend to remain relatively stable. If we decide to delete a #Beta feature, we will typically deprecate it for one release before deleting it.
So it won't happen silently (as far as I observed, usually there's more than one release with deprecating though).
Which brings me the the last point:
On the other hand, if you want something taken out of #Beta, file an issue. We generally promote features out of #Beta only when it's specifically requested, so if you don't ask, it won't happen.
To sum up: I'd suggest you to file a ticket to promote EvictingQueue and make it non-#Beta, which would remove any doubts about it. On the other hand, the EvictingQueue's implementation is quite simple and standalone, so if it's removed (unlikely) you can repakckage it (i.e. use ProGuard) or even copy the code to your project (with all the licenses).
Related
We use ConcurrentLinkedHashMap from https://code.google.com/p/concurrentlinkedhashmap/ in a project and I saw a note that it was integrated into Guava's MapMaker and CacheBuilder back in 2010. The info is very brief:
Integration of the algorithmic techniques into MapMaker will be released in Google Guava r08 and is heavily based on this version.
What does it mean exactly?
The concurrentlinkedhashmap project seems to be still active.
Was it just a one time integration to bootstrap the Guava cache package?
Have the two projects evolved independently since 2010?
If so, what are the main differences between them today?
What does it mean exactly?
Guava is the long term replacement and most of the time you should use it. The history is that ConcurrentLinkedHashMap figured out the algorithms, Guava subsumed it, and then focused on adding features.
The concurrentlinkedhashmap project seems to be still active.
It has always been a weekend project, so active means that I have a scratch to itch or responded to a change request. It is also easier to experiment in CLHM than Guava, so I tended to prove out ideas there prior us porting them over. My involvement with Guava was as a 20%-er.
Was it just a one time integration to bootstrap the Guava cache package?
Yes. We first overhauled MapMaker and then split out caching into a dedicated API. It is a one-way migration of ideas and improvements into Guava.
Have the two projects evolved independently since 2010?
Both have stayed dedicated to their goals. The motivation behind ConcurrentLinkedHashMap was to figure out how to write a truly concurrent cache without taking shortcuts. The goal behind Guava is to provide a feature rich library with a beautiful API and solid implementation for broad usage.
What are the main differences between them today?
Guava is packed with features and has a full time team at Google supporting it. Use it!
ConcurrentLinkedHashMap has higher absolute concurrency by decorating, instead of forking, ConcurrentHashMap. This allows it to be used with ConcurrentHashMapV8, which is based on a new algorithm. CLHM does not relying on segment locks, which improves write performance and allows for maintaining a single LRU chain. I have an experimental branch with the LIRS policy that I hope to someday finish.
The long term hope is that Doug Lea will one day write a cache inspired by our work and teach us a few things in the process.
Update (3/15): Caffeine is a Java 8 rewrite of Guava's cache. It tries to provide the best of ConcurrentLinkedHashMap and Guava, modernized with Java 8, and adopting the techniques that I've learned since those previous projects.
MapMaker maximumSize in google guava library is marked as #Beta. It's a very useful feature to set the maximum size when you use it as a cache and I would like to use it in production code. From experience with other google products beta can be pretty solid. Anyone know why it's a #Beta?
It is used in production at Google and there are no immediate plans on API changes. There is consensus to support weighted entries and we'll likely continue to evolve the algorithm to be closer to ConcurrentLinkedHashMap's. In this case #Beta is just to indicate that the method contract isn't officially set in stone.
From the javadoc:
Signifies that a public API (public class, method or field) is subject to incompatible changes, or even removal, in a future release. An API bearing this annotation is exempt from any compatibility guarantees made by its containing library.
So it doesn't mean it's questionable quality, or unsuitable for production use, they just reserve the right to change it later.
If your project is an end application that is not intended to be embedded in other peoples' applications, then anything with #Beta is totally safe to use. The API may change later and IMO the greatest risk you run is that you might have to change your code a little bit because a method signature changed.
If, on the other hand, you are developing some sort of framework where you expect your project will be on the classpath of many clients, you probably want to avoid #Beta. You never know if your clients will have a different version of Guava on the classpath -- in which case unpredictable things can happen.
ColinD explains this best here: Best Way To Use Guava
edit: Ah yes, and to answer your specific question of why call it #Beta -- it's just to document this fact that MapMaker maximumsize is still subject to change in the next release (It's not that the code is thought to be flaky, unstable, or unsuitable for production)
I was just about to include the HtmlUnit library in a project. I unpacked the zip-file and realised that it had no less than 12 dependencies.
I've always been concerned when it comes to introducing dependencies. I suppose I have to ship all these dependencies together with the application (8.7 mb in this particular case). Should I bother checking for, say, security updates for these libraries? Finally (and most importantly, actually what I'm most concerned about): What if I want to include another library which depends on the same libraries as this library, but with different versions? That is, what if for instance HtmlUnit depends on one version of xalan and another library I need, depends on a different version of xalan?
The task HtmlUnit solves for me could be solved "manually" but that would probably not be as elegant.
Should I be concerned about this? What are the best practices in situations like these?
Edit: I'm interested in the general situation, not particularly involving HtmlUnit. I just used it here as an example as that was my current concern.
Handle your dependencies with care. They can bring you much speed, but can be a pain to maintain down the road. Here are my thoughts:
Use some software to maintain your dependencies. Maven is what I would use for Java to do this. Without it you will very soon loose track of your dependencies.
Remember that the various libraries have different licenses. It is not granted that a given license works for your setting. I work for a software house and we cannot use GPL based libraries in any of the software we ship, as the software we sell are closed source. Similarly we should avoid LGPL as well if we can (This is due to some intricate lawyer reasoning, don't ask me why)
For unit testing I'd say go all out. It is not the end of the world if you have to rewrite your tests in the future. It might even be then that that part of the software is either extremely stable or maybe not even maintained no more. Loosing those is not that big of a deal as you already had a huge gain of gaining speed when you got it.
Some libraries are harder to replace later than others. Some are like a marriage that should last the life of the software, but some other are just tools that are easily replaceable. (Think Spring versus an xml library)
Check out how the community support older versions of the library. Are they supporting older versions? What happens when life continues and you are stuck at a version? Is there an active community or do you have the skill to maintain it yourself?
For how long are your software supposed to last? Is it one year, five year, ten year or beyond? If the software has short time span, then you can use more to get where you are going as it is not that important to be able to keep up with upgrading your libraries.
It could be a serious issue if there isn't a active community which does maintain the libraries on long term. It is ok to use libraries, but to be honest you should care to get the sources and put them into your VCS.
Should I bother checking for, say, security updates for these libraries?
In general, it is probably a good idea to do this. But then so should everyone upstream and downstream of you.
In your particular case, we are talking about test code. If potential security flaws in libraries used only in testing are significant, your downstream users are doing something strange ...
Finally (and most importantly, actually what I'm most concerned about): What if I want to include another library which depends on the same libraries as this library, but with different versions? That is, what if for instance HtmlUnit depends on one version of xalan and another library I need, depends on a different version of xalan?
Ah yes. Assuming that you are building your own classpaths, etc by hand, you need to make a decision about which version of the dependent libraries you should use. It is usually safe to just pick the most recent of the versions used. But if the older version is not backwards incompatible with the new (for your use case) then you've got a problem.
Should I be concerned about this?
IMO, for your particular example (where we are talking about test code), no.
What are the best practices in situations like these?
Use Maven! It explicitly exposes the dependencies to the folks who download your code, making it possible for them to deal with the issue. It also tells you when you've got dependency version conflicts and provides a simple "exclude" mechanism for dealing with it.
Maven also removes the need to create distributions. You publish just your artifacts with references to their dependents. The Maven command then downloads the dependent artifacts from wherever they have been published.
EDIT
Obviously, if you are using HtmlUnit for production code (rather than just tests), then you need to pay more attention to security issues.
A similar thing has happened to me actually.
Two of my dependencies had the same 'transitive' dependency but a different version.
My favorite solution is to avoid "dependency creep" by not including too many dependencies. So, the simplest solution would be to remove the one I need less, or the one I could replace with a simple Util class, etc.
Too bad, it's not always that simple. In unfortunate cases where you actually need both libraries, it is possible to try to sync their versions, i.e. downgrade one of them so that dependency versions match.
In my particular case, I manually edited one of the jars, deleted the older dependency from it, and hoped it would still work with new version loaded from other jar. Luckily, it did (i.e. maintainers of the transitive dependency didn't remove any classes or methods that library used).
Was it ugly - Yes (Yuck!), but it worked.
I try to avoid introducing frivolous dependencies, because it is possible to run into conflicts.
One interesting technique I have seen used to avoid conflicts: renaming a library's package (if its license allows you to -- most BSD-style licenses do.) My favorite example of this is what Sun did when they built Xerces into the JRE as the de-facto JAXP XML parser: they renamed the whole of Xerces from org.apache.xerces to com.sun.org.apache.xerces.internal. Is this technique drastic, time consuming, and hard to maintain? Yes. But it gets the job done, and I think it is an important possible alternative to keep in mind.
Another possibility is -- license terms abided -- copying/renaming single classes or even single methods out of a library.
HtmlUnit can do a lot, though. If you are really using a large portion of its functionality on a lot of varied input data, it would be hard to make a case for spending the large amount of time it would take to re-write the functionality from scratch, or repackage it.
As for the security concerns -- you might weigh the chances of a widely used library having problems, vs. the likelihood of your hand-written less-tested code having some security flaw. Ultimately you are responsible for the security of your programs, though -- so do what you feel you must.
I am using eclipse to develop a web application. Just today I have updated my struts version by changing the JAR file. I am getting warnings at some places that methods are deprecated, but the code is working fine.
I want to know some things
Is it wrong to use Deprecated methods or classes in Java?
What if I don't change any method and run my application with warnings that I have, will it create any performance issue.
1. Is it wrong to use Deprecated methods or classes in Java?
From the definition of deprecated:
A program element annotated #Deprecated is one that programmers are discouraged from using, typically because it is dangerous, or because a better alternative exists.
The method is kept in the API for backward compatibility for an unspecified period of time, and may in future releases be removed. That is, no, it's not wrong, but there is a better way of doing it, which is more robust against API changes.
2. What if I don't change any method and run my application with warnings that I have, will it create any performance issue.
Most likely no. It will continue to work as before the deprecation. The contract of the API method will not change. If some internal data structure changes in favor of a new, better method, there could be a performance impact, but it's quite unlikely.
The funniest deprecation in the Java API, is imo, the FontMetrics.getMaxDecent. Reason for deprecation: Spelling error.
Deprecated. As of JDK version 1.1.1, replaced by getMaxDescent().
You can still use deprecated code without performance being changed, but the whole point of deprecating a method/class is to let users know there's now a better way of using it, and that in a future release the deprecated code is likely to be removed.
Terminology
From the official Sun glossary:
deprecation: Refers to a class, interface, constructor, method or field that is no longer recommended, and may cease to exist in a future version.
From the how-and-when to deprecate guide:
You may have heard the term, "self-deprecating humor," or humor that minimizes the speaker's importance. A deprecated class or method is like that. It is no longer important. It is so unimportant, in fact, that you should no longer use it, since it has been superseded and may cease to exist in the future.
The #Deprecated annotation went a step further and warn of danger:
A program element annotated #Deprecated is one that programmers are discouraged from using, typically because it is dangerous, or because a better alternative exists.
References
java.sun.com Glossary
Language guide/How and When to Deprecate APIs
Annotation Type Deprecated API
Right or wrong?
The question of whether it's right or wrong to use deprecated methods will have to be examined on individual basis. Here are ALL the quotes where the word "deprecated" appears in Effective Java 2nd Edition:
Item 7: Avoid finalizers: The only methods that claim to guarantee finalization are System.runFinalizersOnExit and its evil twin Runtime.runFinalizersOnExit. These methods are fatally flawed and have been deprecated.
Item 66: Synchronize access to shared mutable data: The libraries provide the Thread.stop method, but this method was deprecated long ago because it's inherently unsafe -- its use can result in data corruption.
Item 70: Document thread safety: The System.runFinalizersOnExit method is thread-hostile and has been deprecated.
Item 73: Avoid thread groups: They allow you to apply certain Thread primitives to a bunch of threads at once. Several of these primitives have been deprecated, and the remainder are infrequently used. [...] thread groups are obsolete.
So at least with all of the above methods, it's clearly wrong to use them, at least according to Josh Bloch.
With other methods, you'd have to consider the issues individually, and understand WHY they were deprecated, but generally speaking, when the decision to deprecate is justified, it will tend to lean toward wrong than right to continue using them.
Related questions
Difference between a Deprecated and Legacy API?
Aside from all the excellent responses above I found there is another reason to remove deprecated API calls.
Be researching why a call is deprecated I often find myself learning interesting things about the Java/the API/the Framework. There is often a good reason why a method is being deprecated and understanding these reasons leads to deeper insights.
So from a learning/growing perspective, it is also a worthwhile effort
It certainly doesn't create a performance issue -- deprecated means in the future it's likely that function won't be part of the library anymore, so you should avoid using it in new code and change your old code to stop using it, so you don't run into problems one day when you upgrade struts and find that function is no longer present
It's not wrong, it's just not recommended. It generally means that at this point there is a better way of doing things and you'd do good if you use the new improved way. Some deprecated stuff are really dangerous and should be avoided altogether. The new way can yield better performance than the deprecated one, but it's not always the case.
You may have heard the term, "self-deprecating humor". That is humor that minimizes your importance. A deprecated class or method is like that. It is no longer important. It is so unimportant, in fact, that it should no longer be used at all, as it will probably cease to exist in the future.
Try to avoid it
Generally no, it's not absolutely wrong to use deprecated methods as long as you have a good contingency plan to avoid any problems if/when those methods disappear from the library you're using. With Java API itself this never happens but with just about anything else it means that it's going to be removed. If you specifically plan not to upgrade (although you most likely should in the long run) your software's supporting libraries then there's no problem in using deprecated methods.
No.
Yes, it is wrong.
Deprecated methods or classes will be removed in future versions of Java and should not be used. In each case, there should be an alternative available. Use that.
There are a couple of cases when you have to use a deprecated class or method in order to meet a project goal. In this case, you really have no choice but to use it. Future versions of Java may break that code, but if it's a requirement you have to live with that. It probably isn't the first time you had to do something wrong in order to meet a project requirement, and it certainly won't be the last.
When you upgrade to a new version of Java or some other library, sometimes a method or a class you were using becomes deprecated. Deprecated methods are not supported, but shouldn't produce unexpected results. That doesn't mean that they won't, though, so switch your code ASAP.
The deprecation process is there to make sure that authors have enough time to change their code over from an old API to a new API. Make use of this time. Change your code over ASAP.
It is not wrong, but some of the deprecated methods are removed in the future versions of the software, so you will possibly end up with not working code.
Is it wrong to use Deprecated methods or classes in Java?"
Not wrong as such but it can save you some trouble. Here is an example where it's strongly discouraged to use a deprecated method:
http://java.sun.com/j2se/1.4.2/docs/guide/misc/threadPrimitiveDeprecation.html
Why is Thread.stop deprecated?
Because it is inherently unsafe.
Stopping a thread causes it to unlock
all the monitors that it has locked.
(The monitors are unlocked as the
ThreadDeath exception propagates up
the stack.) If any of the objects
previously protected by these monitors
were in an inconsistent state, other
threads may now view these objects in
an inconsistent state. Such objects
are said to be damaged. When threads
operate on damaged objects, arbitrary
behavior can result. This behavior may
be subtle and difficult to detect, or
it may be pronounced. Unlike other
unchecked exceptions, ThreadDeath
kills threads silently; thus, the user
has no warning that his program may be
corrupted. The corruption can manifest
itself at any time after the actual
damage occurs, even hours or days in
the future.
What if don't change any method and run my application with warnings that I have, will it create any performance issue.
There should be no issues in terms of performance. The standard API is designed to respect some backward compatibility so applications can be gradually adapted to newer versions of Java.
Is it wrong to use Deprecated methods or classes in Java?
It is not "wrong", still working but avoid it as much as possible.
Suppose there is a security vulnerability associated with a method and the developers determine that it is a design flaw. So they may decide to deprecate the method and introduce the new way.
So if you still use the old method, you have a threat. So be aware of the reason to the deprecation and check whether how it affects to you.
what if don't change any method and run my application with warnings that I have, will it create any performance issue.
If the deprecation is due to a performance issue, then you will suffer from a performance issue, otherwise there is no reason to have such a problem. Again would like to point out, be aware of the reason to deprecation.
In Java it's #Deprecated, in C# it's [Obsolete].
I think I prefer C#'s terminology. It just means it's obsolete. You can still use it if you want to, but there's probably a better way.
It's like using Windows 3.1 instead of Windows 7 if you believe that Windows 3.1 is obsolete. You can still use it, but there's probably better features in a future version, plus the future versions will probably be supported - the obsolete one won't be.
Same for Java's #Deprecated - you can still use the method, but at your own risk - in future, it might have better alternatives, and might not even be supported.
If you are using code that is deprecated, it's usually fine, as long as you don't have to upgrade to a newer API - the deprecated code might not exist there. I suggest if you see something that is using deprecated code, to update to use the newer alternatives (this is usually pointed out on the annotation or in a Javadoc deprecated comment).
Edit: And as pointed out by Michael, if the reason for deprecation is due to a flaw in the functionality (or because the functionality should not even exist), then obviously, one shouldn't use the deprecated code.
Of course not - since the whole Java is getting #Deprecated :-) you can feel free to use them for as long as Java lasts. Not going to notice any diff anyway, unless it's something really broken. Meaning - have to read about it and then decide.
In .Net however, when something is declared [Obsolete], go and read about it immediately even if you never used it before - you have about 50% chance that it's more efficient and/or easier to use than replacement :-))
So in general, it can be quite beneficial to be techno-conservative these days, but you have to do your reading chore first.
I feel that deprecated method means; there is an alternate=ive method available which is better in all aspects than existing method. Better to use the good method than existing old method. For backward compatibility, old methods are left as deprecated.
I'm in the process of reviewing a code base (~20K LOC) and trying to determine how to migrating it from 1.4.2 to 5. Obviously, it's not an overnight project and the suggestion which I have received is to write new code against Java 5 and migrate the old code in a piece-meal fashion. Also, I'm no expert in the new features in Java 5 (i.e. I know of them, but have never written any for production use).
My questions:
What features of Java 5 are typically used in production code? (i.e. generics, auto-boxing, etc.) Are there features to be avoided / not considered to be best-practices?
What are the best refactoring strategies which I can use migrate a code base of this size? (i.e. make changes to classes one at a time only when a class is edited, etc.) Objective - reduce risk on the code base. Limitation - resources to do refactoring.
Any advice is appreciated - thanks in advance.
UPDATE - a year too late, but better late than never? =)
Thank you for all of the comments - lots of great points of view. In the life of a software developer, there's always going to be the projects you strive to finish but never get around to because of something more "urgent".
With respect to the use of Java 5 (at that time), it was something which was required in the client's production environment, so that was why we did not use Java 6.
I found that the stronger typing for collections, enums and unboxing of primitives were the features I tend to apply the most, both to old and new code. The refactoring was fairly straight-forward, but code comprehension improved significantly and standards became easier to enforce. The ones I had the most trouble with was the generics; I think it's a concept which I still haven't had a chance to fully grasp and appreciate yet, so it was difficult for me to find previous cases where the application of generics was appropriate.
Thanks again to everyone who contributed to this thread and apologies for the late follow up.
Java 5 is almost completely backwards compatible with Java 4. Typically, the only change you must make when you migrate is to rename any usages of the new enum keyword in the Java 4 code.
The full list of potential compatibility problems is listed here:
http://java.sun.com/j2se/1.5.0/compatibility.html
The only other one that I've run into in practice is related to the change in the JAXP implementation. In our case, it simply meant removing xerces.jar from the classpath.
As far as refactoring goes, I think that migrating your collection classes to use the new strongly-typed generic versions and removing unnecessary casting is a good idea. But as another poster pointed out, changing to generic collections tends to work best if you work in vertical slices. Otherwise, you end up having to add casting to the code to make the generic types compatible with the non-generic types.
Another feature I like to use when I'm migrating code is the #Override annotation. It helps to catch inheritance problems when you're refactoring code.
http://java.sun.com/j2se/1.5.0/docs/api/java/lang/Override.html
The new concurrency library is very useful if your code uses threading. For example, you may be able to replace home-grown thread pools with a ThreadPoolExecutor.
http://java.sun.com/j2se/1.5.0/docs/relnotes/features.html#concurrency
I would definitely take the approach of updating the code as you change it during normal maintenance. Other than the compatibility issues, I don't think there is a compelling reason to use the new Java 5 features unless you're already changing the code for other reasons.
There is one very real issue with the "viral" nature of generics; once you start introducing them at a given layer in an architecture you generally want to introduce it at the layer above & below as well. I have found that introducing generics is probably best done in full "verticals". But you do not have to do all the verticals at once.
This is a really hard question to answer because it depends on what code will be affected and how critical that code is.
First and foremost, when migration is a nontrivial undertaking, do yourself a favour and upgrade to the latest version of Java, which would be Java 6 not Java 5. Java 6 has been out for a year and a half or more and is mature. There's no reason to not pick it over Java 5 (imho).
Secondly, like any software project, your goal should be to get something into production as soon as you possibly can. So you need to identify a slice of your system. The smaller the better, the more non-cdritical, the better.
The other thing to do is just try starting up your app under Java 6 and seeing what breaks. It might be worse than you expected. It might be much better.
The other thing you'll probably need to be aware of is that by the sounds of it you will have jars/libraries in your app that have since been deprecated. Some may not even be compatible with Java beyond 1.4.2. You will probably want to upgrade all of these to the latest version as well.
This will probably mean more stuff breaking but using old/deprecated APIs is just kicking the can down the street and causes you other problems.
There are exceptions to this where upgrading can have far-reaching consequences. Axis1 to Axis2 comes to mind. Those situations require more careful thought.
As for what features are used... all of them pretty much. I can't think of any that should be avoided off the top of my head.
Also, I just noticed the size of your project: ~20K LOC. That's actually quite small (eg I've written an app about that size in the last 3 months by myself).
Lastly, this also depends on how easily you will find things that break. If you have good unit test coverage then great. That's pretty rare though. If you can just run through the app and reliably find problems it's not too bad.
The problematic situations are where scenarios are hard to test and it's likely you won't uncover problems straight away. That calls for more caution.
You would want to migrate stuff that doesn't work in the transition from 1.4 to 5 (not sure what that would be), but I'd be wary of migrating stuff for the sake of it.
If you do take this route, some questions:
Do you have comprehensive test coverage ? If not, you should write unit tests for the code you're going to be migrating.
Do you have components that are widely used within your codebase ? If so, they are probably candidates to be migrated in terms of their API (e.g. using generics etc.)
In terms of what's widely used from Java 5. Generics is important and makes your life a lot easier. I don't see autoboxing too much, nor enums (this is all relative). Varargs almost never. Annotations are useful for frameworks, but I consume these. I don't think I've ever implemented one myself.
20 (non-comment) kloc should be small enough to insert generics with a big bang. Obviously make sure your code compiles an runs on Java SE 5 first. The relatively easy thing about generics is that adding them makes very little change to semantics (certain overloadings can change because of implicit cases - Iterator<char[]> iter; ... System.out.println(iter.next()); as a bad example off the top of my head).
Some cases adding generics will highlight conceptual problems with the code. Using one Map as two maps with disjoint key sets, for example. TreeMap is an example in the Java library where a single class has two distinct mode (using Comparator<T> or Comparable<T>).
Things like enhanced-for and auto-boxing are very local and can be added piecemeal. enums are rarer and might take some thinking about how you are actually going to use them.
I think you're going about this the wrong way. Your plan shouldn't be to update all current code to Java 1.5, your plan should be to ensure that all current code runs exactly the same in 1.5 as it did in 1.4.2, and that all future code written will work fine in 1.5.
I've gone through a few transitions like this of varied sized code bases. The goal was always to make sure we had a ton of unit tests so that we could easily plug in 1.5 and run our tests through it. We actually encountered about 10 problems, mostly related to regular expression libraries not supporting something or supporting something differently.
Write all new code in 1.5 then, and if you change an older class for whatever reason, spend a minute and implement generics, but there's no reason to refactor everything. That sounds a bit dangerous to me if you don't have the tests in place.