Adrián Chaves

I’m addressing the scrapy-zyte-api issue in https://github.com/scrapy-plugins/scrapy-zyte-api/pull/256. For future reference, this was causing the following kind of error in scrapy-zyte-api 0.29.0: ``` crawler = request = , is_start_request = False...

They passed in CI. Maybe you need the latest changes from the main branch (CI automatically merges those into your PR changes before running): `git fetch ; git merge master`

Downloader middlewares’s `process_request()` gets called right before a request is sent. However, if that method returns a request instead of `None`, meaning it changed the request, the old request is...

Even if there were some underlying issue, while I look for time to look into that project to reproduce and understand that, I wonder if setting `RANDOMIZE_DOWNLOAD_DELAY` to `False` may...

I think his point was to enforce a 0.1 minimum delay, i.e. not allow users to set a lower delay. Which I don’t like. We can provide sane defaults, but...

I don’t think we support that for search_dates that the moment. [We do have something like that for parsing, though](https://dateparser.readthedocs.io/en/latest/usage.html#using-datedataparser). As a workaround, for now, you could parse search match...

> it may be easier to reap the task somewhere If that is so, great. However, I personally don’t feel like a 1-year-old version of Twisted is too much to...

I’m not sure if it is a good idea to work on this until https://github.com/scrapy/scrapy/issues/6658, which is in progress but may take some time, has been addressed.

Hmm… If those Windows issues in pinned envs can be solved by upgrading a minimum version of some dep, it might be worth it.

We usually do something like 1, having the old module trigger a deprecation warning, and then we remove the module after 1 year.