An open discussion list for topics related to the eduGAIN interfederation service.

[Davide Vaghetti <davide.vaghetti AT garr.it>]

Hi Ian,

On 16/03/20 11:49, Ian Young wrote:
> > On 2020-03-14, at 14:34, Guy Halse <guy AT tenet.ac.za 
> > <mailto:guy AT tenet.ac.za>> wrote:
> >
> > SAFIRE looks at the size of the logo when we cache it; UKFed looks at 
> > the size of the fully-encoded data: URL. While we were both applying a 
> > nominal "50KiB" cut off, the same 45464 byte unencoded logo was below 
> > my 50KiB threshold and the resulting 60642 byte encoded version was 
> > above theirs.
>
> Good point. While we're talking about the edges of this, I should also 
> clarify that the thresholds the UKf currently has on the table are not 
> kibibytes (KiB, x1024) but kilobytes (kB, x1000). Another potential 
> point of confusion if we're not precise: 40KB, for example, is 
> ambiguous. Alex's post was precise.
>
> I don't have a strong preference for which units we describe the 
> thresholds in, obviously we've gone for the URL because that's the thing 
> we see in metadata. I will note that if people want to standardise the 
> file size instead, you need a way to take the URL prefix (e.g., 
> "*data:*image/png;base64,") into account for filtering unless you want 
> people filtering to convert the URL back to a file to see how big it is.

These are the URL prefix of inline logos I've found in the current 
eduGAIN metadata aggregate [1]:

data:image/gif;base64
data:image/ico;base64
data:image/jpeg;base64
data:image/jpg;base64
data:image/png;base64
data:image/pngl;base64
data:image/vnd.microsoft.icon;base64
data:image/x-icon;base64

Looking only at the image media type the lengths goes from 3 to 18 
characters, or from "gif" to "vnd.microsoft.icon".

In the IANA image Media Type registry the current maximum length of an 
assigned image media type name is 28 characters, though according to 
RFC6838 naming requirements, media types names can add up to 127+1+127 
chars (comprehensive of structured syntax suffix if I understand it 
well). If my calculations are good the upper bound is 255 characters + 
11 ("data:image/") + 8 (";base64,").

Couldn't we just ignore those bytes or am I overlooking something?

Cheers,
Davide

[1] https://mds.edugain.org/edugain-v1.xml
[2] https://www.iana.org/assignments/media-types/media-types.xhtml#image


>      -- Ian

--
Davide Vaghetti
Consortium GARR
Tel: +390502213158
Mobile: +393357779542
Skype: daserzw