royale-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Carlos Rovira <carlosrov...@apache.org>
Subject Re: YUI compressor for CSS (relates to apache/royale-compiler#39)
Date Tue, 10 Apr 2018 06:45:16 GMT
Hi Alex,

ok so, I'll defer this for now, and can get back to it as other things
settle in some months.
I don't see me right now as a release manager, maybe some time in the
future, I'm getting new skills as I progress, for example in the last days
with fixing things in the compiler, but I can't go faster so I can make
things that potentially damage the project.
If I see easy things to do in the meanwhile that minifies content I'll do
so we get half of the path with our custom code.

@Harbs, one more thing I forgot to say before, you must notice is that
Jewel will increase a bit in size in CSS since I'll try layouts to be
mostly CSS rules, in the other hand that will sabe lots of redundant inline
code in HTML setting margins, displays, paddings, and more.

Thanks


2018-04-10 5:47 GMT+02:00 Alex Harui <aharui@adobe.com.invalid>:

> Hi Carlos,
>
> Each thing that YUI can do will make a certain amount of difference.  I
> would expect removing comments and whitespace will make the biggest
> difference.  I thought you said you already had that code ready.  How much
> does that save?
>
> If you want to take on the integration of YUI and be the release manager
> for the release that includes it, that's fine with me.  I'd rather just
> deal with removing comments and whitespace for now.  Maybe next year we'll
> run out of other more important things to do and folks will be demanding
> the remaining optimizations that YUI can do, but right now, I think we
> have bigger fish to fry.
>
> I think we don't have enough people power to do things perfectly.  Folks
> will have to live with things that aren't ideal for a little while longer.
>  I'd much rather spend my time bringing Serkan up to speed, as well as
> Alina's new colleague, getting the emulation components up and running and
> doing and recording a talk on Royale than trying to integrate YUI.  I'll
> bet your changes save 6 of the 8K and the remaining 2K isn't going to make
> a difference among the 100's of kilobytes of JS that have to come down and
> run.
>
> FWIW, I wish there was a lighterweight theme, but maybe it isn't possible.
>  MDL brings down 139K of CSS (minified), The Flat Bootstrap theme is 128K
> (minified).  I'm not worried about that right now.
>
> My 2 cents,
> -Alex
>
> On 4/9/18, 11:32 AM, "carlos.rovira@gmail.com on behalf of Carlos Rovira"
> <carlos.rovira@gmail.com on behalf of carlosrovira@apache.org> wrote:
>
> >Alex,
> >
> >2018-04-09 19:44 GMT+02:00 Alex Harui <aharui@adobe.com.invalid>:
> >
> >> Yes, YUI does lots of things, but how often are folks going to write the
> >> CSS it can optimize?  If you change our emitter so that it removes line
> >> breaks, last semicolon, and most comments, everything else YUI can do
> >>will
> >> likely have minimal effect.
> >>
> >> If YUI is BSD it can be used, but I personally do not want to take the
> >> time to verify that ALL of YUI is BSD compliant, update the builds,
> >> LICENSE, etc, when the net benefit might be that someone was lazy about
> >> margins and YUI saved them 20 bytes.  If that's where you want to spend
> >> your energy, that's fine, but I really think we have more important
> >>things
> >> to do.
> >>
> >
> >If YUI is only 16 java classes of code. Since is a Yahoo library we can
> >expect it is truly BSD right?
> >(I mean is not like to go other library by some individual that will be
> >more tricky)
> >and if BSD is ok for us, I think I'll take much less time importing and
> >using it than looking how it works
> >and try to copy the code and make it work for us.
> >
> >What is more obscure to me is things related to ANT and builds...and maybe
> >I'll need some help there from you
> >
> >But I think this solution is more robust that introducing our own code
> >that
> >not only can introduce other bugs
> >(since can be removing things that should not be removed) and in the end
> >can cost us much time imagining what's
> >happening. So I prefer to depend on well proven and used code than in my
> >own code, that's my humble opinon on this
> >
> >About what we save: I tried it and the results for current JewelTheme are:
> >
> >-debug mode : 31k
> >-YUI: 22k
> >
> >that's 34% of reduction
> >
> >Think that actual theme can be completed at 30%, so the the savings can be
> >significant.
> >
> >
> >
> >
> >--
> >Carlos Rovira
> >https://na01.safelinks.protection.outlook.com/?url=
> http%3A%2F%2Fabout.me%2
> >Fcarlosrovira&data=02%7C01%7Caharui%40adobe.com%
> 7C0dbfc9a79786425477d808d5
> >9e484f43%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%
> 7C636588955758870976&s
> >data=MlTcgPBeX5nGSNzxlg2jbZwOMYgCU1wQnFV1w62ci1E%3D&reserved=0
>
>


-- 
Carlos Rovira
http://about.me/carlosrovira

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message