Related
I often draw a dataflow on a sheet of paper. Even the planning of my little tools is done on a paper.
There exists UML.
The problem is - I don't like it. All the tools I've used (Visio and a lot of online editors) are just not flexible for my hands. With a pencil you can easily draw shapes and connect them, describe them.
What could you suggest in order to create a diagram of data-flow, sequence diagram, etc. in the fastest, most natural and easiest way except on the computer not the paper :)
****Useful links as posted in comments:** SO Link #1 SO Link #2
Right now I am curious about 2 things and one of them was in my minds quite long ago:
1) Mindmap - I've tried a while ago, quite liked it but abandoned. Hoever will give it another try
2) Whiteboard. It would be the easiest and most natural method, except that taking a photo and storing it somewhere on a computer would make the process repetitive and boring.
Has anyone other interesting ideas? I would really like to hear what others are using to design their software and the progress of it.
Thanks a lot!
Why do you want to hand-draw the UML at all whether it's on paper or on the computer?
I agree that you need a model to represent the design. But even in large projects of about 500 man-months, I observed that only 3-4 sequence diagrams really matter and have a chance of surviving the entire lifecycle of the application. Those 3-4 sequence diagrams (and class diagrams that represent their static time relationships), usually represent the high level design of the application.
Or, look at it this way:
Any decent enterprise application will not have 20 different call flows. There will be one or two generic (or abstract) call flows, which all the concrete use cases implement. Let us take a simple Struts / EJB application. The generic flow will be something like - an action class calling a validator and then calling a stateless session bean, which in turn calls a domain class, which will call a DAO. All the use cases of the application just implement this flow with concrete classes that are specific to that use case.
Do you agree?
If you do not, I would like to hear about applications that have 20 different call flows and survived for 5 years after the first release.
If you agree with me, we are boiling down to 3-4 class and sequence diagrams even for large enterprise applications comprising several thousand classes. Why is it a big deal how you draw and maintain those 3-4 diagrams?
You might say that you want to document all the use cases for training or documentation purposes. During my last 14 years of experience in the real enterprise software world, I don’t remember seeing well 'maintained' UML documentation. First of all, good documents are difficult to produce and are not found that often. Secondly, they are out of sync with the code most of the time. Most of my experience is with large banks, insurance companies, Auto companies, etc. Those environments are just too hectic and their resources are limited (really? Are we talking banks? Yes, difficult to believe, but true) for 'maintaining' good documentation.
So am I suggesting that we get rid of UML?
No. We need visual models to represent complex systems. Human brains seem to be at their best when processing visuals. The visual cortex, which is responsible for processing the visual images, is the largest system in the human brain.
So what is a reasonable solution to easily produce and maintain UML models?
Probably we are better off using the current crop of UML tools to draw those 3-4 high-level UML diagrams. If you hate using them, check option 3 below.
For the diagrams at the next level of abstraction (any useful models should have different levels of abstraction), generate the UML from source code. You can generate both class and sequence diagrams.
In this age of agile methodologies, why not just write the shell classes and generate those 3-4 high-level UML class and sequence diagrams as well? This way there won't be any UML to maintain at all.
The source code is the truth.
Can you argue against that statement? If not, why not generate the models from the source code itself? I am not suggesting the round-trip engineering, by the way. I am just suggesting a one way trip - from code to models.
There are 2 main problems however with the generated UML.
When we hand draw a class diagram, we show the relations between the classes involved in a scenario. Most existing class diagram generating tools allow the user to drop the Java classes (the source code) into the tool and the tool automatically shows the relations between the classes. The problem here is, how does one know about the classes involved in a scenario to begin with?
The second problem is the verboseness of the generated diagrams. There are tools available to generate runtime sequence and class diagrams for a scenario. But the diagrams are often very verbose and defeat the purpose of models, whose purpose is to highlight the important aspects and filter out unimportant details.
Good UML generating tools should address both the above problems. There are a few tools in the Java domain that try to address these problems. Check the discussions below:
What tools should I use to visualize structure of my code
Are there any tools for detecting architectural and design patterns in code?
I hope I answered the original question:
Has anyone other interesting ideas? I would really like to hear what others are using to
design their software and the progress of it.
I am the author of the runtime UML generating tool MaintainJ, but I tried to address the original question in an objective manner. Your comments are welcome.
There are various tools that allow you to create diagrams based on textual input. There's some up-front learning in that you need to learn the syntax. However it's not hard to do. Once you have, creating diagrams can be very fast. There are some downsides; in most cases there's limited ability to change the layout/style. Significance of that will depend on whether you like their style or not.
There's a growing number, here's a few you might want to look at:
UMLet: desktop app, supports most UML plus various other diagrams. Can also create your own custom shapes & connectors. FOSS.
WebSequenceDiagrams.com: online sequence diagrams.
TextUML: desktop app. Focus is executable models, auto-generates class diagrams. FOSS. It also has an online commercial sibling.
hth.
I like using a whiteboard and a camera. For even more flexibility, use post-it notes on the whiteboard.
I use ER diagrams (on the whiteboard) to model my data, and message sequence charts (on the whiteboard) to model the data flow. I'll also do quick mockups of UI pages on the whiteboard.
Asides from that, I use Ruby/Rails to code server side and HTML/CSS/jQuery/JS on the client.
If even Visio is not flexible enough, I'd suggest a digital whiteboard or touchscreen with a whiteboard software. After some accommodation you could probably use a simple tablet (without display) as well - they are really cheap.
Regarding pure software: we are trying to achieve a "pen-like" input method with UML Lab, but it currently supports Class Diagrams only...
I think that the UML and code should be mixed using a class diagram. You model your architecture with the class diagram (e.g package, classes etc....) then you code your business finally multiple iterations between code and model.
I think that UML should more be oriented to code but not to focus on textual input.
The problem with standard languages, such as UML, is that you have to invest a considerable effort to learn the language and the modeling tools. These languages are defined by an expert consortium, e.g. OMG, that proposes a language specification suited to the biggest overlap of design problems in a certain domain.
Why not defining your own language that fits exactly to your needs and your specific problem? Such languages are termed Domain-Specific Languages (DSL). Instead of investing into learning a language that's complex, you invest into the definition of a languages that exactly suits your needs.
There are numerous approaches that support the definition of DSLs. The most widespread is the Generic Eclipse Modeling System (GEMS). Personally, I made great experience with GrGen due to its versatility and the possibility to automate working steps using graph transformation.
No. There are various other ways. UML is just an option.
Pen and Paper Prototyping is a great option too, it doesn't have to follow UML.
Mind Map is another great way.
For more adaptive software processes, UML use is encouraged to be as minimum as possible. Such as, teams that practice Agile or XP tend to use UML less and they would rather rely more on informal means to conceptualize the software. In a rigid structured company, UML can be rigorously followed.
I might have a silly question about DDD:
Are there any disadvatages of DDD generraly? I mean, besides of using it when it is not necessary, or needed. (e.g. small/not complex projets)
Thanks
It is very easy to do it wrong.
Eric Evans has said in a JavaOne presentation that DDD is best applicable when there is a lot of business domain complexity. Moreover, he explicitly states that DDD is not suitable for problems where the is substantial technical complexity but that has little business domain complexity. An example of the later is an embedded system with very simple inputs (possibly independent of the number of states it possesses) while simultaneously presenting a lot of technical complexity (in terms of getting the hardware to work.)
How we go about quantifying a lot or a little, that's an open subject. But the narrative should work as a guideline of when and where DDD is best applicable.
-- edit --
I got the video presentation through Emule, and I've never been able to find that same lecture on youtube or similar video venues.
I found this discussion of DDD in the Microsoft Application Architecture Guide to be helpful in understanding the challenges of that particular style:
As the core of the software is the
domain model, which is a direct
projection of this shared language, it
allows the team to quickly find gaps
in the software by analyzing the
language around it. The creation of a
common language is not merely an
exercise in accepting information from
the domain experts and applying it.
Quite often, communication problems
within development teams are due not
only to misunderstanding the language
of the domain, but also due to the
fact that the domain's language is
itself ambiguous. The Domain Driven
Design process holds the goal not only
of implementing the language being
used, but also improving and refining
the language of the domain. This in
turn benefits the software being
built, since the model is a direct
projection of the domain language.
In order to help maintain the model as
a pure and helpful language construct,
you must typically implement a great
deal of isolation and encapsulation
within the domain model. Consequently,
a system based on Domain Driven Design
can come at a relatively high cost.
While Domain Driven Design provides
many technical benefits, such as
maintainability, it should be applied
only to complex domains where the
model and the linguistic processes
provide clear benefits in the
communication of complex information,
and in the formulation of a common
understanding of the domain.
At an italian conference, I talked about the topic (see these slides).
DDD projects require domain experts that are often expensive to hire, since they hold valuable knowledge (read, if you don't need a domain expert, you don't need DDD).
Moreover DDD requires strong skills on the modeler side. In particular they have to be:
Humble, since they have to accept their ignorance and learn from an expert
Really skilled in OOP
Diligent, since they have to track decisions
Comunicative, since they have to explain the domain to the other devs (even through documentation)
Finding such developers can be much more expensive than expected, since you can really know that they are good at the job after some months of trials.
Martin Fowler in his "PoEA" book has a great discussion of when and what domain logic patterns to use.
For instance, simplest pattern, Transaction Script, involves no domain model.
I have seen UML mentioned several places in the last few years, but never had a compelling reason to use it at work so far. Is there any value in putting the time and effort to learn it? (I am a Ruby on Rails developer.)
Edit: I'm also looking for stories about how UML has made a difference in your projects to give me an idea of how I might use it.
Yes. It is the international standard methodology for systems analysis and design. When moving from the initial analysis phase through the project life-cycle, the UML gives a good road-map on where to go and how you got there. A few benefits:
It shows requirements in a way that clients / customers / management can understand.
You can iterate the model and not lose track of where the model was before
Shows exactly how you got from point A to point B
It's a standard, so anyone who knows UML will be able to interpret your diagrams
The "quickest" way to share ideas without needing to go through the entire code-base
No, save your time and don't learn it. If you are a coder and work alone and are a very careful person, you can completely ignore UML.
But, if you work with other people and want to share the result of your work, UML is a Unified language, a lingua franca that even non technical customers understand (at a certain level).
Yes and no.
Yes, learn the basics. Have a quick look at at the different diagram types etc. and have a general idea. That will help you eliminate your hesitance when someone boasts about UML.
No, if your work doesn't require to work with UML, you don't necessarily need to know UML to write good applications. If you need to discuss something on a white board just draw a few circles. That'll do.
It only helped me to produce documentation to management to impress them a little. Other than that I only find the deployment diagrams somewhat useful, that's all.
If you look at UML as a merely graphical language you consider just a very small part of it. You should instead look at UML as the primary language of the Model Driven Architecture framework which is very well implemented in Eclipse.
In Eclipse UML is (correctly) considered as a Platform Independent language with a solid syntax defined by the MOF (ECORE in Eclipse) and semantics defined by the UML specification (http://www.omg.org/spec/UML).
In Eclipse there are good implementation of the two main languages defined by the OMG for Model To Text and Model to Model transformation specifications which are respectively MOFM2T (ACCELEO in Eclipse) and QVT.
Also Eclipse provides an implementation of the Object Constraint Language (OCL) which is used by both QVT and MOFM2T to evaluate queries on models.
All this means you can easily define your own Model to Text and Model to Model transformations , in the form of Eclipse plugin, turning your UML models into whatever you need.
Now I use that to automatically generate thousands of LOC, documentation and tests with an impressive return on investment.
However I know the majority of people doesn't even know UML is a language but think it is just about tiny pictures.
Look at this links for some simple example
http://lowcoupling.com/post/46522537374/the-model-driven-architecture
http://lowcoupling.com/post/47800863669/qvt-in-place-transformations
http://lowcoupling.com/post/47347056110/models-to-text-transformations-with-mofm2t-and-acceleo
I think it's a question of scale in two dimensions: size of problem and size of team.
When a design gets to a certain size diagrams become useful in two ways: first, they help you reason about design issues. second they help you communciate the design to other poeople.
So if the team is say 20 or 30 it really does help to have some clear documentation of pieces of the overall design.
Personally I use UML maybe four or five times a year, but when I need it, I really need it. It really is better to ase a standard diagramming technique that devise your own. And with good tools it's pretty painless.
I would say that I use only a small subset of UML, class diagrams and occasional collaboration diagrams.
I'm looking for a good slideshow/pdf/video explaining the differences in approach and thinking from hand-written threading of applications compared to the more abstracted and easier to use message passing and actor models. Does anyone know of existing resources to explain these concepts with good diagrams and visualizations?
It is slightly difficult to make direct comparisons without long, painful digressions, or a largely theoretical discussion. However, the following can be easily read, and I believe that the comparisons will form naturally for anyone familiar with the threading model.
Google's language GO uses message passing among co-routines as a core part of its concurrency model. There is a lot more information on GO at golang.org, and the following URL provides information about their concurrency model in GO.
http://golang.org/doc/effective_go.html#concurrency
This is a paper written by Edward Lee (Berkeley EECS department chair) called The Problem with Threads. It is a pitch for the actor model, and is a good read. Also note that there are other papers by Edward that deal with the problem with threads (visit his homepage for more papers).
http://www.eecs.berkeley.edu/Pubs/TechRpts/2006/EECS-2006-1.pdf
UML is a great language to model software for business requirements, but there is a growing community that points some disadvantages for some lacking features.
What are the most significant disadvantages that you find crucial for UML and what could it be a good alternative to solve this lacking features?
The biggest one is that it's yet another layer of red tape that gets in the way of just $#%$#% coding the thing and making it work.
The fact that people use it to "model software for business requirements", as you put it, and other such process-oriented claptrap. UML started out as a conventionalised way for programmers to communicate software to other programmers in a pictorial form. In that sense it's just formalised napkin-scribbling - and as such it is very effective. You can draw a UML class diagram on a whiteboard and I can understand it without quibbling over notation.
But somewhere along the line someone got the idea that a drawing notation could somehow be a process in it's own right, or at least a formal part of a larger process. And that's just silly. UML diagrams are a fine way to illustrate books, and quite useful as a means for engineers to scribble ideas back and forth. But that's where it should have ended.
I can say at least three:
It takes a lot of time to keep the diagram reasonable and synchronized with the actual code. UML diagrams don't run, but require a lot of time. So they are good only if your organization size can manage them
You cannot represent every condition in a sequence diagram. It's impossible if you want to deliver. So state diagrams should convey basic facts, not all the possible outcomes.
Good UML software costs money and it takes some time to master properly.
So, I think UML is good as a complementary documentation role, and only if the size of your organization allows it.
Solutions... well, in the end, diagramming is just a way to convey high level information to another person, in space or time (e.g. could be you in some year time). Extreme Programming shifts the burden of information retrieval from dead tree to living brain. Of course, it assumes that the living brain never forgets, and never quits. Extreme programming uses redundancy to reduce the impact of such occurrences. In a large company, a strong layoff round could wipeout entire teams, so storing information into brains can be risky. On the other hand, large companies have human power to waste, hence the diagramming.
Also, as WDuffy points out, if you are a designer, and you have to communicate to a team of programmers what they have to implement, it's much easier to use a UML diagram. Of course, a small company with a small team has generally small goals, and you can organize people with a different style. A small company UMLing will only produce UML diagrams of their revolutionary product, and then it will be bankrupt.
UML is not good nor bad. It can be a good tool, but it must be used in the proper context.
Lacking features?
well, I found that UML is strongly aimed at an Object Oriented vision of the world. Our company mainly developed in python, with a strong focus on module level routines. Objects were lightweight data containers, but all the logic was done at the module level. It's difficult to properly model this implementation style at the UML level, unless you resort to some "hacks" in the terminology. I guess it's difficult to model in UML for functional or procedural languages.
Another thing I find annoying is the assumption of use case modeling as a diagram. My experience is that the best way to convey a use case is to write a short story or a short code tingling the feature you want to convey. The story should be short, one page maximum.
This approach has two advantages: if your story is a written prose, the Q/A team can read and test it easily. If your story is code, you can put it as a functional test and let it run during the night. A diagram does not satisfy any of these value added needs.
One issue with UML is due to its universality: things in UML cannot be always implemented directly in the target language, or some languages have capabilities that cannot be expressed in UML. So it can be better to know the implementation language beforehand, which restrain its universality.
See also the criticisms section on UML wikipedia page:
Standards bloat
Problems in learning and adopting
Cumulative Impedance/Impedance Mismatching
Dysfunctional interchange format
It's not Agile
What should have been the last word on UML was written by frustrated student "Candide Smith", well, really Eiffel author Bertrand Meyer.
Another disadvantage of UML is that it tends to overemphasize design, which can lead to 'analysis paralysis' (people over-analyze their problem) and feature creep (loosing sight of the actual problem). A UML design can only take you so far in solving a problem, and you have to be careful to jump into the code soon enough (but not sooner ;-).
UML is somewhat less applicable to the brave new world of loose typing and NoSQL databases. It has OO ideas of Class as a data structure rather than classification embed in it.
Another disadvantage, although not self-inflicted, is that it doesn't explicitly to facilitate abstraction. Everyone I know uses UML tools for more abstract modelling, but the way standards are written that is not obvious.
Another problem with UML (and big design up front in general) is that it's sometimes hard to anticipate all the nitty gritty implementation problems that you'll run into that may affect your design until you actually start implementing something. Granted, I'm a bioinformatics research programmer that works on small one-man projects, but I don't even believe in any design up front, at least for small projects. I believe in the following:
Make it work. (Just get a prototype up and running that has all the basic functionality, no matter how much it sucks. This forces you to see all the little nitty gritty details that might not come through in a formal analysis. Having an actual implementation of your idea makes it easier to see whether the idea was even really worth doing in the first place or whether it should be scrapped altogether.)
Make it right. (Only now, when you have a working prototype and you know that all the nitty gritty implementation problems are at least in principle solvable do you worry about good design. Refactor the heck out of it to follow good programming practices, reduce coupling, do proper error handling, yada yada yada.)
Make it fast. If it's application code, you'd better have proof that you've found the slow part. If it's generic library code, you'd better have good reason to believe that the piece of code could reasonably be the slow part in some use case for the library, i.e. don't optimize a function that noone would ever call in a loop.
For Class Diagrams in UML it only makes sense to use them if there is automated way to generate code directly from diagram. I have implemented such UML editor tool based on 4-tiers meta levels recommended by OMG (Object Management Group) and we had great success using UML in team of 5 devs over 2 years doing around 20-30 architectural iterations. The diagram was the root artifact of automated build chain making impact on hundreds of derived artifacts, APIs, generated Docs, DDLs, projects, tests etc.
So by itself UML in Class Diagrams part is great "programming" language if you actually do programming in it.
For Class Diagrams in UML if it is not translatable in automated way, then its fail.