[HN Gopher] Automatically filing issues when tracked file conten...
___________________________________________________________________
Automatically filing issues when tracked file content changes
Author : simonw
Score : 10 points
Date : 2022-04-28 17:29 UTC (2 days ago)
(HTM) web link (simonwillison.net)
(TXT) w3m dump (simonwillison.net)
| JNRowe wrote:
| Slightly tangential to this post, but related to the series.
|
| Have you had any pushback from the sources you've been scraping
| with actions? I can't imagine how lightly hitting GitHub from
| this action is a problem, but a few of the other scrapers you've
| discussed are hitting resources in ways we've definitely seen
| website operators get upset about before.
|
| I ask because I've been playing with a couple of scrapers for
| info I'm interested in, but I'm a little reluctant to toggle the
| repo visibility switch. Partly, because I don't want the grief
| and selfishly because I don't want the source to disappear.
| simonw wrote:
| Nothing at all so far.
|
| I've been pretty careful to be polite though - the majority of
| my scrapers work no more than once an hour, and the few I've
| done that are more frequent than that don't run more than once
| every ten minutes (GitHub Actions can't be scheduled more often
| than that).
|
| Those tend to be sites which I would expect to be able to
| handle the traffic very easily.
|
| When I'm operating sites myself I don't tend to even notice a
| scraper that only hits a single page every few minutes. The
| things that hurt me are crawlers that hit every page on a site,
| potentially more than a request per second and with traffic
| patterns that are certain to hit un-cached sources.
| JNRowe wrote:
| Excellent, thanks!
|
| I'm purposely being careful too. _Big_ backoff on any error,
| curl 's --time-cond from git commit timestamp, etag where
| possible, etc.
|
| I'm not particularly concerned about an actual DoS'ing issue,
| more being on the end of one of _those_ stories that pops up
| every so often( "website owner sics lawyers on F12").
| simonw wrote:
| Oh interesting - I hadn't considered using --time-cond.
|
| I'm always a bit suspicious of If-Modified-Since because
| I'm never sure which timezones will be in effect - so I
| tend to use ETags instead, which achieve the same goal
| without involving any clock time.
|
| I actually wrote a tiny Python CLI tool for doing those,
| designed to be used in CI:
| https://pypi.org/project/conditional-get/
___________________________________________________________________
(page generated 2022-04-30 23:01 UTC)