Vanessa, steps were discovered that replaced our long
Posted: Wed Jan 22, 2025 10:31 am
An example of an auxiliary script that has become willingly used is filling an object according to a table. But there was another problem with export scenarios - due to their complexity and branches, they take a very long time to execute. As a natural result, testing began to take many hours, which became very critical. Sweat They tried to solve the problem by parallelizing the execution of tests in several "threads", but then a number of technical problems arose that could not be solved. For example, one of the main problems was licensing 1C testing clients in docker containers. In parallel, they tried to launch several testing clients on different virtual desktops, but then another problem arose: the child windows of the testing client could open on a different virtual desktop, which led to the tests crashing.
Also, in this mode of operation, even mexico phone number list on a fairly powerful server, there were problems with drawing interface objects of 1C itself. Sometimes these objects were drawn a few seconds after the form was opened, which also did not contribute to the stable operation of the tests. I don't know about my colleagues, but at this stage I had a crisis, and my eye started twitching at the mere mention of tests. But somehow, once again, looking through the list of changes in the new version of export scenarios. Having revived my spirit, as a check, the next failed tests were rewritten using these steps and, lo and behold - the tests work quickly and do not fail! The only thing left to do was to rewrite all the existing tests for the new version of Vanessa.
A decent amount of time was devoted to this activity. But you can't abandon product development and devote all your time to rebuilding your infrastructure. Therefore, at the next meeting, it was decided that since tests began to fail much less frequently and their execution time was quite within acceptable limits, it was possible to slightly reduce the rate of their rewriting. Tests Although the work on rebuilding the testing infrastructure is not yet finished, we can already draw some conclusions. Let's list all the "rakes" that our team has been successfully dancing on for quite a long time: test scripts should not rely on data generated by other test scripts export scripts are nice but they should be simple without complex branches and such if you need complex steps that are not in the Vanessa library, then it is better to replace the export script with your own steps (instructions for creating these steps will be below) some technical aspects of testing in 1C remain either unresolved or have a “specific” appearance It remains to tell how to actually implement these "magic" custom steps for Vanessa.
Also, in this mode of operation, even mexico phone number list on a fairly powerful server, there were problems with drawing interface objects of 1C itself. Sometimes these objects were drawn a few seconds after the form was opened, which also did not contribute to the stable operation of the tests. I don't know about my colleagues, but at this stage I had a crisis, and my eye started twitching at the mere mention of tests. But somehow, once again, looking through the list of changes in the new version of export scenarios. Having revived my spirit, as a check, the next failed tests were rewritten using these steps and, lo and behold - the tests work quickly and do not fail! The only thing left to do was to rewrite all the existing tests for the new version of Vanessa.
A decent amount of time was devoted to this activity. But you can't abandon product development and devote all your time to rebuilding your infrastructure. Therefore, at the next meeting, it was decided that since tests began to fail much less frequently and their execution time was quite within acceptable limits, it was possible to slightly reduce the rate of their rewriting. Tests Although the work on rebuilding the testing infrastructure is not yet finished, we can already draw some conclusions. Let's list all the "rakes" that our team has been successfully dancing on for quite a long time: test scripts should not rely on data generated by other test scripts export scripts are nice but they should be simple without complex branches and such if you need complex steps that are not in the Vanessa library, then it is better to replace the export script with your own steps (instructions for creating these steps will be below) some technical aspects of testing in 1C remain either unresolved or have a “specific” appearance It remains to tell how to actually implement these "magic" custom steps for Vanessa.