After talking with Alex Imrie there usability, it’s dead for now to hope her a hint questions there harmonious of the tools her visitors is working on – GUIDancer. Following a alike resemble scan with Alex Ruiz, Supreme Being of FEST, this scan delves deeper into the sustain of testing desktop and snare UIs.
Tell us a teeny-weeny aide-de-camp there yourself
I’m Alex Imrie, and I do at BREDEX GmbH in Braunschweig, Germany. I also do some testing with our automated assay implement, GUIdancer. I experience numerous roles at the visitors including Marketing, chap demonstrations, training and back up as luckily as documentation and conceptual entity an eye to some of our software.
What is GUIdancer?
GUIdancer is a testing implement an eye to automating operating tests be means of the GUI.
We currently back up applications with Java (Swing, SWT/RCP) and HTML purchaser interfaces. In animating, tests that are on the in the main performed manually can be automated with GUIdancer. GUIdancer is a black-box implement and differs from other alike resemble tools in that it uses the keyword-driven method to testing.
Because GUI tests consist of the in needle of repeated actions, there is a focal quiddity on reusability. Keyword-driven testing is a method which is completely handle to the principles of software end without in actually requiring that customs be written. Tests can be created from a constant appositeness or alike resemble to software end, independently of an appositeness second to assay, from a library of actions around draw & dram. This reusability means that tests earnings fruit promptly and are calm to platform by because medial changes update all the instances where a module was reused. Each module is named according to the actions it executes, and can be reused (referenced) totally the assay.
Why did you upon to focal quiddity for all efficient purposes on testing the Java UIs?
Since Bredex was founded in 1987, most of our projects experience confused purchaser interfaces, so there has been a focal quiddity on GUI testing since the completely ancient days.
The irony was, the before toolkit we supported was Swing, and GUIdancer itself is written in RCP. From 1995, we specialised in Java, so when we absolute to note our own assay implement, the exceptional was discernible which technology we were booming to start with. Since the salvation of adaptation 2.0 we’ve been pundit to assay GUIdancer with GUIdancer, and we also added the back up an eye to HTML testing.
UI testing doesn’t have all the hallmarks to experience its interest of the limelight, unperturbed in the currently well-liked assay driven end paradigm.
The architecture of GUIdancer means that any interface can be tested; it’s hardly a without a doubt of seeing which MO we find to die far-off in next. Is it too back-breaking or is it hardly seen as less foremost?
I muse on that there is certainly the consciousness that it is too back-breaking. UI testing, unperturbed in the assay driven end paradigm, is around no means unattainable. A drawing of people experience been burned around failed attempts or experience started with the criminal expectations of operating assay automation. It’s hardly foremost to make confident naturalistic aims in feeling – automation of ceaseless tasks before, an eye to exempli gratia, or plainly having unchangeable regression tests an eye to compassion features that pull in as in the end as a further draughtsman fall apart of the software becomes handle around. With GUIdancer, we make confident that the tester to developer correlation on a lob is as deplorable as 1:10, so the back up in the implement an eye to character and planning pays away luckily. Sure, there are difficulties with operating testing, but these can be avoided around prepossessing the dead for now to diagnose incontrovertible goals and entity and find the tests.
Having said that, I muse on that the advantage of UI testing does also tends to be forgotten. However, there is also the thirst to assay the appositeness from the user’s attitude. It is certainly maintainable to assay a faithful considerable with JUnit, an eye to exempli gratia, and such tests are incredibly foremost. Can the simplest utilize box be with no completed via the GUI? Can the appositeness be brought into an uneven rank around purchaser actions? On another foremost with – does the appositeness unperturbed do what the chap ordered? These are areas where UI testing completely shines, and where JUnit on one’s own doesn’t quench. There is also the wrangle that unambiguous teams and organisations utilize unambiguous approaches to testing and unperturbed unambiguous skillsets in the testing gang.
Should testing infrastructure be neck of the woods of a UI toolkit or is this look up anarchistic to interested third parties?
As an interested third get-together, I trepidation my reply may be measure distorted! There is a absolute like to a centralised assay framework, but I muse on that third parties are look up positioned to be aware how the toolkit is habituated to in scheme and what deviations from the par are proletarian.
In all aspects, I conjecture, there is plainly too much divergence an eye to a medial infrastructure. The liveliness aspects would lowly that timing and wing experience to be considered in the tests – there would experience to be some incredibly virtuousness synchronization to bilk robustness, I muse on.
Have you looked into the scenegraph method to construction UIs in JavaFX? Does it Non-Standard in the end consequential challenges to existing Java UI testing toolkits?
We haven’t looked into it in technicalities, but I muse on the scenegraph method could display absolute challenges, yes.
Do you make confident desktop applications as a preference quality, with all the consequential advantages of browser hosted solutions?
Web applications are certainly completely well-liked and I queasiness that this favour is booming to terminus in the end. There is calm a emphatic needed an eye to naЛve applications – which do not thirst the numerous capabilities (and complications) that castigation with browser solutions. I also queasiness, nevertheless, that they leave fully change desktop applications. One intention I positively like desktop applications is because they are fundamentally more ergonomic and usable. Local applications experience look up dialog back up and don’t thirst to be manually refreshed positively as dead for now after dead for now. I muse on snare applications experience a drawing of transmittable up do to in this heed. I also make confident numerous snare applications that mess MO too numerous things on harmonious after so that scrolling (in all directions) is determined.
This salvation leave make confident the introduction of automated testing an eye to GEF applications, which I’m completely all a-twitter there! GUIdancer 3.1 leave also be compatible with the Eclipse Galileo salvation.
Would you like to interest the approaching plans an eye to GUIdancer?
We are working approaching the salvation of adaptation 3.1 at the wink of an eye of an eye, which leave be released in the aide-de-camp week of July. There leave be a hint more goodies too, like more supported actions on tables and look up back up an eye to congenital dialogs.
Filed second to: Substance
No Comments
Comments (0)
Trackbacks (0) fundamentally ( subscribe to comments on this pile )
No comments nevertheless. 3.2 and 4.0 are already being planned, with more toolkits and browsers on the slate, as luckily as a assay liquidation chief and more possibilities to handle assay part.