mirrors/git-annex
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

analysis

Browse source
This commit is contained in:
Joey Hess 2013-09-28 11:27:52 -04:00
parent 1ba3165135
commit bd431408b5
1 changed files with 30 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

30
doc/bugs/importfeed_fails_when_using_the_option_--lazy_for_specific_podcast.mdwn
View file

@ -41,3 +41,33 @@ git-annex importfeed --fast http://schoolsucksproject.com/category/podcast/feed/
unable to access url: http://schoolsucks.podomatic.com/enclosure/2013-09-16T10_10_06-07_00.mp3 unable to access url: http://schoolsucks.podomatic.com/enclosure/2013-09-16T10_10_06-07_00.mp3
failed failed
... ...
> (There is no --lazy option. You seem to mean --fast.)
>
> This fine web server rejects the User-Agent used by curl:
<pre>
joey@darkstar:~>curl http://schoolsucks.podomatic.com/enclosure/2013-09-18T20_40_40-07_00.mp3
Forbidden
joey@darkstar:~>wget http://schoolsucks.podomatic.com/enclosure/2013-09-18T20_40_40-07_00.mp3
--2013-09-28 11:18:36-- http://schoolsucks.podomatic.com/enclosure/2013-09-18T20_40_40-07_00.mp3
Resolving schoolsucks.podomatic.com (schoolsucks.podomatic.com)... 38.99.42.46, 38.110.155.212
Connecting to schoolsucks.podomatic.com (schoolsucks.podomatic.com)|38.99.42.46|:80... connected.
HTTP request sent, awaiting response... 200 OK
<pre>
> git-annex always uses curl for checking file sizes. So the workaround
> is to use `git annex addurl --relaxed` on this url, which will skip
> the size check. However, if you only had curl installed, `git-annex get`
> would again try to use curl to get the file, and would still fail.
> It only happens to successfully download because git-annex chose
> to use wget and this site has apparently forgotten to block that.
>
> I don't know if it makes sense for git-annex to vary the user-agent
> to get around such (incredibly stupid) blocking. It could retry
> with a random user-agent, but that could be construed as abusive
> behavior; this site has asked us to go away. The only choices
> that seem really defensible would be to add a --user-agent
> switch, and/or to make git-annex set a default user agent header
> of "git-annex", rather than relying on the curl/wget defaults.
> --[[Joey]]