Linus' idea of "good taste" code

Patrick Schluter via Digitalmars-d digitalmars-d at puremagic.com
Wed Nov 2 23:32:07 PDT 2016


On Thursday, 3 November 2016 at 06:11:08 UTC, Joakim wrote:
> On Sunday, 30 October 2016 at 10:04:02 UTC, Patrick Schluter 
> wrote:
>> On Sunday, 30 October 2016 at 06:39:42 UTC, Joakim wrote:
>>>
>>> It is not worth it, the web is dying.  I was stunned to see 
>>> this chart of mobile web usage in the US:
>>>
>>> https://mobile.twitter.com/asymco/status/777915894659964928
>>>
>>> This isn't some third-world country with mostly 2G usage, the 
>>> web numbers in those places are much worse. Combined with 
>>> mobile passing even TV for time spent, there is no point in 
>>> wasting time porting D to a dying platform.
>>
>> Yes, because outside of web on mobile nothing else exists... 
>> bwahahahah
>
> Pretty soon it won't:
>
> https://mobile.twitter.com/asymco/status/793401867053195264
>

Even that chart shows a flattening to an asymptote not a linear 
trend. This means desktop will maube go a little bit down but it 
won't disappear. What people often forget is that professional 
office PC will never be completely replaced by mobile. What also 
happens in that branch (i.e. in office environment) is still a 
continuation to replace PC application by browser applications. 
This means that still focussing on good web solutions server or 
client side is a good investment in any case.

> On Sunday, 30 October 2016 at 16:35:54 UTC, Laeeth Isharc wrote:
>> And wouldn't the changes to runtime and phobos be quite 
>> similar for both dmd and ldc? I don't see how the work flow 
>> would be any different as a language user whether you used an 
>> LDC with wasm back end,  or dmd with similar.
>
> The changes to druntime and phobos wouldn't depend on the 
> compiler used, but it is difficult to test unless you have a 
> compiler with working codegen, so that usually comes first.  
> You can go ahead and make changes to druntime- not much has to 
> be done for phobos, as the idea is to encapsulate 
> platform-specific code in druntime, though a minority of phobos 
> does call platform-specific APIs- based on the spec or 
> available headers, but you won't know if it will work well till 
> you can run it.
>
>> Joakim - native on mobile is so much better (setting aside 
>> having to deal with Apple or Google)   but I guess the browser 
>> isn't going away on the desktop for a while yet.
>
> I'm actually a heavy web user, have been for almost a 
> quarter-century (though I don't use webapps, mostly reading), 
> which is why that chart was so surprising to me.  While native 
> mobile apps are usually more responsive, they are not ideal for 
> reading, as I'm not going to install and load up The Verge's 
> app, or an app for every other news site, every time.
>
> The problem for the desktop browser is that the desktop is 
> going away, as the linked tweet above shows.

No it is not. Linear extrapolation of an incomplete chart is 
almost always erroneous.

> I went from using a FreeBSD desktop and a dumbphone five years 
> ago to an Android smartphone and two Android tablets today, ie 
> no desktop or laptop since my ultrabook died late last year.  
> In my household, we went from using two smartphones, two PC 
> laptops, and a Mac laptop four years ago to three smartphones, 
> three Android tablets, and a Mac laptop today.
>
> This is a shift that is happening in most households, as a PC 
> overserves most and a mobile device will do.  Many D users are 
> power users who cling to old tech like the desktop and the web, 
> so they are missing this massive wave going on right now.  I 
> myself missed the death of the mobile web, as I'm such a heavy 
> user.

still bwahahaha, web technology will stay a bit longer, panicking 
is a bit premature yet.


More information about the Digitalmars-d mailing list