Compression is even more important today on mobile networks: mobile apps
are very verbose over the net, and you can easily pay the extra volume. In
addition, mobile networks are frequently much slower than what they are
advertized, even if you pay the extra subscription to get 3G/4G, you depend
on antennas and the number of peoples around you.
In my home, 3G/4G in faact does not work at all, and this is the case in
many places around in my city, even though they are sold to have full
coverage (for example, just downloading an application or updating it is
simply impossible: I have to be at home connected to my Wifi router, but
when its internet link fails (this happens sometimes for several hours, I
have extremely slow connections on 3G/4G (which is also overcrowded at the
same time, and only delivers 2G speeds).
Lot of people have to support frequently low bandwidths on mobile networks,
independantly of the price they paid for their subscription.
So compressing data is stil lextremely important (even for texts or for the
smallest web requests). Thanks, compression is now part of the web
transport, but this does not mean that apps must learn to represent their
interchanged data efficiently, and develop less verbose protocols and APIs).
There are more and more people using mobile networks now than fixed
landline internet accesses (or home wifi routers connected to it, and even
for them, fiber access is still jsut for a minority of people in dense
areas, the others don't get more than an handful of mebgatit/s on their DSL
access: if you look at worldwide internet connections a large majority of
people don't get more than 2 megabit/s: this is enough for reading/sending
SMS or phone calls, or exchanging emails, but not if you need frequent
updates to your apps and your apps are too verbose and there are too many
apps in the background: many people cannot view videos on their mobile
access, or only with very poor quality if they view it "live" (they cannot
also download them slowly due to lack of storage space on their mobile
device, so videos have to remain short in total volume and duration).
So I disagree: compression is absolutely needed (even more today than iut
was in the past when mobile Internet accesses were still for a minority.
Mobile networks are not really faster today (their bandwidth does not
double every three year like local performances of devices ! But with this
extra local performance, you can support more complex compression schemes
that require more CPU/GPU power which is no longer a bottleneck, when the
real bottleneck is the effectively available bandwidth of the mobile
network (smaller than the connection bandwidth because this bandwidth is
shared... and expensive).
2015-06-03 15:24 GMT+02:00 David Starner <prosfilaes_at_gmail.com>:
> Chris wrote:
> > There is no way to compare 2 HTML elements and know they are talking
> about the same character
>
> That's because character identity is a hard problem. Is the emoji TIGER
> the same as TONY THE TIGER or as TONY THE TIGER GIVING THE VICTORY SIGN?
>
>
> http://www.engadget.com/2014/04/30/you-may-be-accidentally-sending-friends-a-hairy-heart-emoji/
>
> Note that even in Unicode, the set ẛ ᷥ ſ ṡ s S Ŝ may be considered the
> same character or up to seven different characters, depending on
> case-folding, canonization and accent dropping.
>
> > Similarly, there is no way to search or index html elements. If a HTML
> document contained an image of a particular custom character, there would
> be no way to ask google or whatever to find all the documents with that
> character. Different documents would represent it differently.
>
> You can index links to images. If two documents represent it differently,
> then I go back to the above; we can't know that they're the same thing.
>
> On Tue, Jun 2, 2015 at 7:11 PM Chris <idou747_at_gmail.com> wrote:
>
>> You can’t ask the entire computing universe to compress everything all
>> the time.
>
>
> Anytime we care about how much space text takes up, it should be
> compressed. It compresses very well. On the other hand, it's rare that
> anyone cares anymore; what's a few hundred kilobytes between friends?
>
Received on Wed Jun 03 2015 - 09:06:41 CDT
This archive was generated by hypermail 2.2.0 : Wed Jun 03 2015 - 09:06:41 CDT