6 ms·

Microsoft Playwright: Automate Chromium, WebKit and Firefox

43 points by tomduncalf 7y ago

benologist 7y ago
This looks like an amazing next-step for Puppeteer, and it addresses some of the annoyances too. I generally use a loop that will wait-and-retry when commands fail as it seems like virtually anything can crash in the right circumstances. > click waits for the element to be available and visible by default > With Playwright, we encourage setTimeout-free automation
gsnedders 7y ago
As the FAQ says, this is written by largely the same team as who worked on Puppeteer, who largely moved to MS last year. That said, I have severe reservations about their claims that it's cross-browser: They mention WebDriver, and how they hope to "steer it towards the bi-directional communication channel", but they've not at any point tried to engage with the working group, and they've historically actively pushed back against any attempt to standardise CDP or anything very similar to it before (despite CDP already having an ecosystem around it). They were the owners of the CDP protocol within the Chrome team and it was their lack of support for any standardisation that is the reason why there is no standardised protocol today. They say they support Firefox, but this is support in the same sense as puppeteer-firefox: it relies on a large patch to Firefox, which looks similar to the one written for puppeteer-firefox, and which has no chance of getting landed. It seems like they haven't taken the feedback on-board from when they tried this with puppeteer-firefox: springing a large patch onto the Mozilla community is unlikely going to get accepted, especially when they haven't addressed the architectural concerns raised previously. They could've engaged with the work that has been happening around Firefox with a duplex testing protocol over the past year, but they haven't, and they've written their own thing in isolation. The fact that Firefox now already has another protocol partly implemented makes it even less likely that their patch and their new protocol will get accepted. The group now maintaining Puppeteer at Google is actively engaging with those working on WebDriver and actively engaging with the work happening around Firefox which makes it far more likely that Puppeteer in the future will support an unmodified Firefox binary.
cryptos 7y ago
How is the performance compared to Cypress (which is limited to Chromium)?

43 points by tomduncalf 7y ago

This looks like an amazing next-step for Puppeteer, and it addresses some of the annoyances too. I generally use a loop that will wait-and-retry when commands fail as it seems like virtually anything can crash in the right circumstances. > click waits for the element to be available and visible by default > With Playwright, we encourage setTimeout-free automation

gsnedders 7y ago

As the FAQ says, this is written by largely the same team as who worked on Puppeteer, who largely moved to MS last year. That said, I have severe reservations about their claims that it's cross-browser: They mention WebDriver, and how they hope to "steer it towards the bi-directional communication channel", but they've not at any point tried to engage with the working group, and they've historically actively pushed back against any attempt to standardise CDP or anything very similar to it before (despite CDP already having an ecosystem around it). They were the owners of the CDP protocol within the Chrome team and it was their lack of support for any standardisation that is the reason why there is no standardised protocol today. They say they support Firefox, but this is support in the same sense as puppeteer-firefox: it relies on a large patch to Firefox, which looks similar to the one written for puppeteer-firefox, and which has no chance of getting landed. It seems like they haven't taken the feedback on-board from when they tried this with puppeteer-firefox: springing a large patch onto the Mozilla community is unlikely going to get accepted, especially when they haven't addressed the architectural concerns raised previously. They could've engaged with the work that has been happening around Firefox with a duplex testing protocol over the past year, but they haven't, and they've written their own thing in isolation. The fact that Firefox now already has another protocol partly implemented makes it even less likely that their patch and their new protocol will get accepted. The group now maintaining Puppeteer at Google is actively engaging with those working on WebDriver and actively engaging with the work happening around Firefox which makes it far more likely that Puppeteer in the future will support an unmodified Firefox binary.

cryptos 7y ago

How is the performance compared to Cypress (which is limited to Chromium)?