Jueves, 3 Mayo 2007

Durante la mañana del jueves se han iniciado las Jornadas de Testeo de Software con la instalación de los stands de los expositores; se ha realizado el registro de asistentes y se han iniciado las jornadas con la bienvenida por parte de Tanja Vos. A partir de ahí se han desarrollado dos sesiones:

  • Rex Black: Quality Risk Analysis (dividida en dos partes)
  • Michiel Vroom: TMap Next, for result driven testing 
Asistentes a las jornadas JTS 2007

Rex Black: Quality Risk Analysis

Durante la primera charla de la jornada, Rex Black se ha centrado en el análisis de riesgos relacionado con el testeo de software. Ha destacado que este análisis en las primeras etapas de desarrollo es importante, ya que habitualmente las empresas testean más o menos en función del tiempo disponible; por eso es importante empezar el testeo por los problemas con mayor riesgo. ¡Desarrollar software es una actividad peligrosa!

Rex Black, durante su exposición

Además de proporcionar algunas definiciones de calidad y riesgo de calidad, habla de tres técnicas para hacer ese análisis de riesgos: estudiar los requerimientos del sistema (no muy útil, según Rex Black); brainstormings; y entrevistas one-to-one.

El objetivo es un esquema en el que se identifican los posibles problemas, asignando a cada uno un impacto y una probabilidad de que ocurra. Ambos parámetros permiten priorizar los riesgos, asignando un nivel importancia (y un tipo de testeo) a cada uno de ellos. Un peligro inevitable (sobre todo para los menos experimentados) de este análisis es ser demasiado detallado.

Asistentes a las jornadas durante una de las ponencias

Rex Black deja claro que, para él, la característica del "risk based testing" es que no tiene como objetivo simplemente encontrar bugs (un objetivo técnico), sino reducir el riesgo. Algunos de los desafíos del análisis de riesgos es conseguir un consenso entre diferentes personas, asignar prioridades (evitando que todo tenga prioridad máxima), asignar riesgos de modo racional, dedicar el tiempo suficiente, etc.

Por último, Rex Black ha mostrado algunos ejemplos concretos del resultado de un análisis de riesgos (en formato de hoja de cálculo), y ha dejado una recomendación: empezar a realizar estos análisis de modo informal y sencillo, haciéndolo de modo más detallado según se adquiere más experiencia.

Michiel Vroom: TMap Next, for result driven testing

Michiel Vroom, de SOGETI

Michiel Vroom, de SOGETI, ha presentado la revisión de su metodología de testeo TMap que está explicada en su libro TMap Next. Con esa revisión han pretendido, a partir de sus experiencias, actualizar TMap, haciéndola más práctica, más integrada con el proceso de desarrollo, considerando el testeo como una actividad económica.

El cliente es el responsable y decisor en el proceso de testeo. Él especifica sus objetivos; a partir de ahí, se determinan riesgos, se calcula el testeo necesario y se estima el coste para el cliente, que decidirá si está de acuerdo o no. A partir del acuerdo se especifican y ejecutan los tests.

Tal como resaltó Rex Black en la ponencia anterior, el análisis de riesgos es muy importante para especificar qué y cómo se debe testear. Como metáfora, usa las cuotas de seguros de accidente: a mayor probabilidad de accidente, mayor coste asociado. Nuevamente recuerda que el cliente es el que marca los objetivos, a partir de los cuales se especificarán las técnicas concretas de testeo a utilizar.

De modo similar a Rex Black, Michiel Vroom aconseja no empezar aplicando la metodología TMap completa en cada caso, sino adaptarla a la situación específica, escogiendo las técnicas útiles y aplicables.

16 comentarios sobre “JTS 2007: jornada matinal del jueves, 3 de mayo”

  1. Hamza dijo:

    Patry, en cualquier tneida de informe1tica te informare1n.Un router+switch con wireless por 801.11g y un receptor que soporte la norma 801.11g (en un usb, tarjeta pci o integrado en el equipo) suele ser suficiente, pues funciona a 54Mbits/segundo. Si tienes problemas de sef1al puedes poner un router 801.11n y un receptor con la misma norma, o si no quieres gastar tanto puedes comprar un receptor usb con una antena de buena sensibilidad y ganancia. A mayor ganancia en dB mejor sere1 la recepcif3n, pero si te alejas mucho, necesitare1s alimentacif3n externa para emitir.Puedes en foros especializados si fuese este el caso.Mucha suerte y un saludo.Nacho AvilesP.D.- El test me da 130 Mbitsps en descarga. Y tengo contratado 30 Mb. Raro, raro, raro .

  2. cialis dijo:

    Hi Huib,I quiet agree with your reasoning that there is no actual necessity to have test managers as a separate role. I much prefer to have the following more general division of craftsmanship in IT:Junior Architect / Designer / Developer / Tester / Etc.Intermediate Architect / Designer / Developer / Tester / Etc.Senior Architect / Designer / Developer / Tester / Etc.Expert Architect / Designer / Developer / Tester / Etc.With each level the tasks they perform would intensify, change or grow in number.And with each level the skills needed to do so would increase and be of a different level accordingly. All in all the tasks needed to accomplish a project in a specific context would be fulfilled, as necessary, by the rolls and skill levels needed.So testers would follow a line of personal development and career more like any other role in IT. More so their behaviour would be similar also. So they would no longer write plans and reports none of the other IT roles do (Test Plan, Test Report, Bug Report, etc.). Reporting such content should be just as much a team effort as the actual development of software. This by the way contradicts your Immaturity section where you, in spite of the tenor of you post, still assign such specific tasks to testers. To conclude: I aggree but I would go a step further and integrate testers into IT completely. Something which by the way goes against the grain of many test methodologies, but thats a different story…Regards,Arborosa

  3. viagra dijo:

    the opposite of fun is not professionalism.I understand it is hard to know when your are done. Can you explain what you mean by “measurable completion”? But do you know when you are done with a (what you call) structured method like TMap or ISTQB? You know when all the test you came up with upfront are complete. But how do you know that you have found all the important bugs? Also: please read by Michael Bolton or by Markus Gärtner about structured and exploratory testing.Let me try to answer your questions:Q: How will those new people on “more practical” course know when they have finished?A: Good question. Let me compare the learning I described with the learning from the TMap course. In the TMap course you see 350 slides and do 4 practical exercises using test design techniques. But you do not test real software. So after 3 days you know a lot about definitions, processes, artifacts, but not how to test real software. In my experience a lot of people who did the TMap course had difficulties getting up to speed after the course. When you do the way of learning I described, I think after three days you tested a lot of things and you learned how to actually test something. Sure in both scenarios the student is far from done. Compare it by the way kids learn. Did you send your kids to a 3 day course in walking school? No, they learned by trying, failing, trying again (and a lot of exploring). Q: What tests did they do?A: Using charters, sessions, logging and debriefs as in SBTM () a tester is very capable of knowing/reporting what they did.Q: What did they forget?A: Do you know what you forget in your testing? I hope you do not claim that you don’t forget anything. Because in that case we would have bug free software…Q: What defect types did they target?A: It depends. But each session can target a different type. In the testing I have observed a lot of TMap and ISTQB trained people do, it struck me how un-targeted they where testing. I have seen a lot of confirmatory testing, but haven’t seen much targeting of specific defects. I don’t think a method will improve this much. Practice and thinking while testing will. My observation is that the more testers use “structured” methods like TMap, the less they think.Q: Which ones did they not look for?A: Same answer as before: do you know which you didn’t look for? And is this method specific or just because you think?Q: What is the risk to the system?A: Who says ET doesn’t do risk based testing? I would claim ET is highly risk based. ET tries to find the most important bugs first. Those are the highest risk, aren’t they? So I do not see why this would be a problem.The questions you ask, give me the feeling that you do not know what exploratory testing is and how it is practiced well. ISTQB and TMap still consider ET as a technique. I think there is much more to it and I would describe it as a mindset and an approach. I invite you to read some of the stuff collected on the . Let me add a last thing to this: me and my team had the pleasure to do exploratory sessions with Michael Bolton. We explored exploratory testing. While trying stuff we were discovering how it works. Every now and then Michael would do a short lecture to explain or illustrate some things. I feel we learned more because we weren’t told, we (in a way of speaking) found it out ourselves. As Confucius said: “Tell me and I forget. Show me and I remember. Let me do and I understand.”

  4. accredited online colleges dijo:

    Rik,You keep seeing some misunderstandings about TMap. Which misunderstandings are you talking about here?I wouldn’t call this context-driven because you are using the business case as a starting point not call this context-driven because of that. Maybe you should read more about context-driven, context-aware and context-specific .There is no book (yet). But a good start would be “” by Kaner, Bach and Pettichord.This Jedi stuff and bosses who do not like people asking questions hasn’t anything to do with testing directly but with culture! I consider these bosses as fake-managers. Testers who choose “something to keep their boss happy are consider fake-testers to me.Jan Jaap Cannegieter raised the question if a tester should only master one approach or method. You say it is very good if testers know more than one. To me it is ESSENTIAL to be considered as a real tester.Exploratory testing is not-well described in the book, but it is also fairly misunderstood by the writers. And they are not alone. ET is misunderstood by the majority of the test population. So everybody who truly understands ET, will not call it a technique.Why is TMap (or the people who use it) trying so hard to be complete? To be the one and only approach/method? Rik do you truly believe that TMap can be used in /any/ situation?Context-driven isn’t the best approach in any situation. Let me give you 3 examples (which can be found in the slides of the presentation James and Michael did for TestNet):* When someone else is entirely responsible for the quality of your work.* When you’re working in a single, specific, well-established and unchanging context (e.g a car-factory).* When your goal is to change the context.Why is regression testing not about exploration? Even ISTQB acknowledges the pesticide paradox. And because (fake-)managers find something hard to plan, they hate it? Testing is hard to plan, any testing is! But we like to think that testing is planable. That is why testers and their fake-managers like test cases counting and planning. Because it gives them the (fake) feeling of being in control…There probably will be examples of CDT used in the wrong way. But does that make it right?IIari uses a definition of “test” that is in his opinion the definition used in the factory school. The definition of “testing” in TMap NEXT is: “Testing is a process that provides insight into, and advice on, quality and the related risks.” So it is a process. Not just test cases. And the process provides insight. In quality and risks. To show whether the problem is solved well enough.You say that TMap is a process. And that might be the problem: it is too much considered as a process.Context-driven is not (only) an approach, it is much more than that! It is a paradigm, an approach and a community. Read this made by Michael Bolton and James Bach. It was used to prepare the TestNet presentation and is published on the . You might also want to read the on the same website.I fully agree with you here Rik: TMap is a valuable toolbox that can be used in a context where it’s suited. But I think we disagree about the how often that context will be found. How much can be changed to still call an approach TMap? I think TMap (as described in the book) is only efficient and effective useable in maybe 10% of all situations. I all other we need to adapt so much, that I can’t call it TMap anymore.I will be happy to help you find your way in the context-driven world I think you are a great tester, your only downside is that you still think TMap can be used in any context… But you are learning. Welcome to the !

  5. levitra dijo:

    That’s not just the best answer. It’s the bestest answer!

  6. cialis goedkoop kopen dijo:

    That’s a sensible answer to a challenging question

  7. generic for levitra dijo:

    No more s***. All posts of this quality from now on

  8. buy viagra online dijo:

    How neat! Is it really this simple? You make it look easy.

  9. cheap auto insurance dijo:

    Reading this makes my decisions easier than taking candy from a baby.

  10. carroll insurance dijo:

    There’s nothing like the relief of finding what you’re looking for.

  11. cialis perscription dijo:

    This introduces a pleasingly rational point of view.

  12. levitra dijo:

    I suppose that sounds and smells just about right.

  13. Home owners insurance California dijo:

    Ya learn something new everyday. It’s true I guess!

  14. car insurance quotes dijo:

    Now I know who the brainy one is, I’ll keep looking for your posts.

  15. auto insurance quote in ny dijo:

    YMMD with that answer! TX

  16. levitra dijo:

    I have been so bewildered in the past but now it all makes sense!

Deje un comentario

*
Teclea la palabra del recuadro, por favor, para comprobar que eres una persona y no un proceso de spam. << Anti-Spam Image



Blog del grupo SQUaC. ¿Conoces nuestros servicios? testeo de software; etc.

SQUaC, una web del ITI