Relative xlink path for GBIF logo breaks in hypothesis.is proxy

18612
Reporter: rdmpage
Type: Feedback
Summary: Relative xlink path for GBIF logo breaks in hypothesis.is proxy
Priority: Unassessed
Resolution: WontFix
Status: Closed
Created: 2016-06-25 15:11:44.26
Updated: 2017-10-10 16:05:06.78
Resolved: 2017-10-10 16:05:06.759
        
Description: I'm hoping this wins a prize for most obscure bug.

I'm using https://hypothes.is to annotate web pages (including scientific articles). This tool looks like it may be emerging as the best ay to annotate the web. One way it works is by sending a web page through a proxy server so that it can be annotated, e.g. https://via.hypothes.is/http://www.gbif.org/species/7753250 (there is also a Chrome extension).

When I do this the GBIF logo is replaced by the "missing image" symbol. I think this is because the following code has a relative URL "/img/header/GBIF-2015-full.svg" rather than an absolute URL:


          
        

This doesn't survive being viewed via a HTTP proxy (if the URL was absolute t would). Could this be fixed?]]>
    

Author: hoefft
Created: 2016-06-27 11:18:29.447
Updated: 2016-06-27 11:18:29.447
        
It isn't a bug so no prize.

I do not know the service - but it shouldn't be done on our side I would say. It is hypothes.is that only does a halfway job of rewriting urls. The logo on GBIF is an svg, and their search/replace only does works for script tags, style tags, images it seems. Other tags is seemingly ignored.

That doesn't mean that we couldn't be accommodating if we see this service being used a lot and hypothes'is do not fix it. But I would report it to them instead/as well as it is a general issue with their product.

We could of course also just change it - but it seems a bit silly and wrong.
    

Author: hoefft
Comment: Do you know if it is used a lot? I must admit i do not know it
Created: 2016-06-27 11:19:18.402
Updated: 2016-06-27 11:19:18.402

Author: rdmpage
Created: 2016-06-27 12:07:23.55
Updated: 2016-06-27 12:07:23.55
        
[~hoefft] I'll talk to hypothesis.is about the bug.

Regarding how widely is it used, they've been under the radar a little but have recently announced a deal with 40 scholarly publishers where those publishers will be incorporating hypothesis into their publishing platforms see https://hypothes.is/annotating-all-knowledge/ So we may end up with most of science publishing being open to annotation.

It's always hard to predict these things, but my guess is that if there is going to be a successful attempt to annotate the web then https://hypothes.is/ is it. Hence I would argue that GBIF should be looking at https://hypothes.is/ closely (and not wasting too much time on the inevitable home-grown tools from the biodiversity informatics community).

This has some implications for site like GBIF, such as ensuring that any site redesign tries to avoid breaking annotations (the hypothesis.is folks have designed some pretty clever ways to make annotation location robust to some changes).

I realise this conversation is much bigger than a comment on an obscure "bug" in JIRA so I'll send something to [~dhobern@gbif.org] and [~trobertson@gbif.org] to make the case that this is something we need to be looking at.
    

Author: rdmpage
Comment: [~hoefft] I've opened an issue on hypothesis.is https://github.com/hypothesis/h/issues/3549
Created: 2016-06-27 12:18:02.4
Updated: 2016-06-27 12:18:02.4

Author: hoefft
Comment: Sounds interesting - I like the idea of it. It is hopefully and probably a forgiving platform - but might require some minor tweaks on our side -be it image urls or custom markup.
Created: 2016-06-27 12:58:01.728
Updated: 2016-06-27 12:58:32.876

Author: hoefft
Comment: unless we decide to promote integration with this service I will close this issue. 
Created: 2017-10-10 16:05:06.777
Updated: 2017-10-10 16:05:06.777