- From: James Graham <james@hoppipolla.co.uk>
- Date: Wed, 21 Aug 2013 18:49:52 +0100
- To: public-test-infra@w3.org
On 21/08/13 11:03, Tobie Langel wrote: > On Monday, August 19, 2013 at 2:03 PM, David Burns wrote: >> By putting the details all in the test html that excludes out of >> process tests, like WebDriver tests, from getting the information >> so we would need to run tests, process information and then abort >> or complete which is not ideal. > > Agreed, but at the end of the day, there is some tradeoff we have to > make between ease of authoring/revieweing and ease of running the > tests. > > Currently I feel like the former is the bottleneck and the one we > should optimize for. > > That said, there should be a number of ways to mitigate the issues > you mention. > > jgraham suggested automating manifest generation and including these > manifests in the repo itself. FWIW since I need something now, I hacked together [1]. It is slightly specific to my needs and obviously could be improved to actually parse html files and identify reftests by something other than filename. But it does allow a per-directory manifest in the format I suggested to be autogenerated (most of the time), and overridden (in special cases). [1] https://github.com/jgraham/html-testsuite/blob/manifest/tools/scripts/update_manifest.py
Received on Wednesday, 21 August 2013 17:50:18 UTC