[Dachs-support] Two questions on datalinking
Markus Demleitner
msdemlei at ari.uni-heidelberg.de
Mon Aug 30 09:43:49 CEST 2021
Hi Yan,
On Thu, Aug 26, 2021 at 02:57:09PM +0200, Yan Grange wrote:
> Thanks Markus and Yannick! Very insightful. My first off the bat reaction to
> the 1e8 limit would be "but we are radio! 1e8 is like... Nothing". Our
> images are about 4.5 times that. So I need to decide how stubborn I am by
> interpreting "~1e8" as an order of magnitude :). Maybe pushing tooling to
> support data links is actually a good action on the RadioIG since I cannot
> imagine this problem to be unique to our data releases :).
For datasets of half a Gig, I'd suggest having datalinks as product
URLs would be a good thing if you can do fast cutouts (which may not
be the case if the data is remote). Without a reasonable cutout
facility, it probably doesn't matter either way.
> As things go I have an unrelated question:
> Right now our accrefs contain +-signs, so getdata URL will also contain a +
> sign. I have noticed dachs to complain if I have +-es in file names. Is
> there a way that people replace plusses in file names? I can not immagine
> that we are the first ones to have this problem given general source naming
> conventions in astronomy (on the Northern hemisphere) :).
A bit of background on this: There's nothing wrong per se with pluses
in URLs, except that ancient compontents sometimes replace them with
blanks (u+0020) when processing them. DaCHS' warning against them
mainly results from half-broken reverse proxies that would have
butchered such links (oh, and there's quote_plus and unquote_plus in
python's urllib.parse that will also do the wrong this with those).
I'd hope in modern environments you probably don't need to bother.
See
http://docs.g-vo.org/DaCHS/howDoI.html#force-the-software-to-accept-weird-file-names
for the recommended way to make DaCHS relax.
-- Markus
More information about the Dachs-support
mailing list