Because, this isn't an issue of performance. We give our users choices and a toolkit to replicate ANY HTML tree they want to. We are proposing to replace what is to me the most obvious Flex-like component name for injecting a span into the DOM. Label is a single-line sequence of text. So is Span.
I'm not seeing a huge performance difference. If you think it is significant enough that our users should be alerted to use SizeableLabel or whatever we call the Div-wrapping-text version, start blogging it after you validate your test case. I noticed that the number of Spans the test appears to report are higher than the number of Divs, but that might be a false alarm. You can be a hero in the web world if you are right and change the advice I see elsewhere on the net to use Span for inline elements and Div for other things. Even the JSPerf page is using spans. Tell them to replace them. Meanwhile, I'm going back to try to migrate an app. Sure wish we had more people helping on that. -Alex We can certainly give them advice that SizedLabel is faster than Label if that's important, but the different didn't seem that significant. On 3/15/18, 6:14 PM, "[email protected] on behalf of OmPrakash Muppirala" <[email protected] on behalf of [email protected]> wrote: >If creating a div is more performant than creating a span, why insist on >using a span? How is that pay as you go? > >Thanks, >Om > >On Thu, Mar 15, 2018 at 5:19 PM, Alex Harui <[email protected]> >wrote: > >> A Flex Label, and any Flex component was a swiss-army knife of features >> and not PAYG. I have not given up on PAYG. I think it will pay off >>when >> we get to the last mile. If we don't stick to PAYG, I think we will >>just >> repeat the history of Flex and produce a fat framework that can't run >> everywhere. Yes, a Div would probably better mimic a Flex Label, but >>the >> Basic label is not a swiss-army knife. >> >> This article is old, but I think it says you can't put a Div in a Span: >> >>https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fdoctype.c >>om%2Fspan-contain-div-html5&data=02%7C01%7Caharui%40adobe.com%7C77b57b271 >>ef7423114c208d58ae0d810%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C6365 >>67620658308491&sdata=o9msOjZpbwywflSfgJDebUdMM3hgNPdN%2Btn%2BrcSZsV4%3D&r >>eserved=0 >> >> I think you can if you change the span's display to "block" but I think >> that is essentially replacing the element. >> >> Everything, including width and truncation should be opt-in, PAYG. >>Make a >> LabelWithWidth or SizeableLabel component that is a div if you want. >> >> Again, the goal is to encapsulate common patterns. I am not seeing a >>lot >> of Divs wrapping text nodes. I see lots of spans wrapping text nodes. >> Basic Label is that pattern. The Basic component set would be >>incomplete >> if there wasn't a component that could inject a Span into the DOM. >>Create >> another component that is a Div wrapping a text node, give it a >>different >> name, and let's move on. >> >> Alina used Label quite often without setting width. No need for those >> instances to be a Div, AFAICT. >> >> Do we even need to discuss span vs div if the truncation bead will work >>by >> setting display:block? >> >> We have a huge opportunity here to get another app migrated and prove to >> the world that Royale is the right choice. But we keep discussing stuff >> that IMO, isn't really moving the big pieces forward. But, IMO, we >>should >> not compromise on PAYG philosophy and patterns. >> >> My 2 cents, >> -Alex >> >> On 3/15/18, 4:27 PM, "[email protected] on behalf of OmPrakash >>Muppirala" >> <[email protected] on behalf of [email protected]> wrote: >> >> >FWIW, it seems like creating divs is faster than creating spans. >> >https://na01.safelinks.protection.outlook.com/?url= >> https%3A%2F%2Fjsperf.co >> >m%2Fspan-vs-div-creation&data=02%7C01%7Caharui%40adobe.com >> %7C3664f690b3da4 >> >e4aefab08d58acc7bf0%7Cfa7b1b5a7b34438794aed2c178de >> cee1%7C0%7C0%7C636567533 >> >215453966&sdata=oPPp2e9aRbQ%2FngKbcFMHMCmXi%2BKa%2B% >> 2F0aA98XFFqxhFA%3D&res >> >erved=0 >> > >> >Thanks, >> >Om >> > >> >On Thu, Mar 15, 2018 at 3:38 PM, Greg Dove <[email protected]> wrote: >> > >> >> Observing from the sidelines, fwiw I agree with Harbs here. Span to >>me >> >> makes more sense as part of the text value of a label and not as the >> >>basis >> >> for the representation of the Label itself - div seems closer to >>what I >> >>was >> >> used to with legacy Flex (which may no longer be the 'reference' for >> >> choosing an implementation). >> >> I don't think there is any semantic or other reason that makes span >>the >> >> logical default tag for Label, in fact I recall that being a problem >> >>many >> >> times with some work I was involved with last year using FlexJS. >> >> But maybe I am missing something in terms of understanding. >> >> >> >> >> >> >> >> >> >> On Fri, Mar 16, 2018 at 11:08 AM, Harbs <[email protected]> >>wrote: >> >> >> >> > I’m suggesting to change it because span does not fit the use case >>of >> >> > Label. Width and truncate are just two applications which indicate >> >>that >> >> > span is not the correct element. I’m pretty sure I had other edge >> >>cases >> >> > which were likely effected by the fact that Label is a span, >>although >> >>I’m >> >> > not positive. >> >> > >> >> > I disagree with your distinction between span and div. >> >> > >> >> > Both span and div can have children. Spans can contain divs and >>divs >> >>can >> >> > contain spans. At its core, Span has nothing to do with text. It’s >> >>just >> >> > often used for text because text (within its container) is inline. >> >> > >> >> > MDN says[1] "<span> is very much like a <div> element, but <div> >>is a >> >> > block-level element whereas a <span> is an inline element.” >> >> > >> >> > Label is *not* necessarily an inline element. We have all kinds of >> >> layouts >> >> > for Label. Many (most?) of them are not inline. >> >> > >> >> > If you feel really strongly about this, I’ll revert even though I >> >> disagree. >> >> > >> >> > FWIW, the truncate bead does not need to change the element type. >>It >> >>can >> >> > explicitly set the display of the element to block. >> >> > >> >> > Harbs >> >> > >> >> > >> >>[1]https://na01.safelinks.protection.outlook.com/?url= >> https%3A%2F%2Fdevel >> >>oper.mozilla.org%2Fen-US%2Fdocs%2FWeb%2FHTML%2FElement% >> 2Fspan&data=02%7C0 >> >>1%7Caharui%40adobe.com%7C3664f690b3da4e4aefab08d58acc >> 7bf0%7Cfa7b1b5a7b344 >> >>38794aed2c178decee1%7C0%7C0%7C636567533215453966&sdata= >> xrtF9H7%2BjFzqGDax >> >>nj0%2BuY3hsHu6BaUb4Yp15ucwhyc%3D&reserved=0 < >> >> > >> >>https://na01.safelinks.protection.outlook.com/?url= >> https%3A%2F%2Fdevelope >> >>r.mozilla.org%2Fen-US%2Fdocs%2FWeb%2FHTML%2FElement% >> 2Fspan&data=02%7C01%7 >> >>Caharui%40adobe.com%7C3664f690b3da4e4aefab08d58acc >> 7bf0%7Cfa7b1b5a7b344387 >> >>94aed2c178decee1%7C0%7C0%7C636567533215453966&sdata= >> xrtF9H7%2BjFzqGDaxnj0 >> >>%2BuY3hsHu6BaUb4Yp15ucwhyc%3D&reserved=0> >> >> > >> >> > > On Mar 15, 2018, at 11:25 PM, Alex Harui >><[email protected]> >> >> > wrote: >> >> > > >> >> > > Label in Flex and Royale is not intended to have any children >>(other >> >> than >> >> > > bold, italic and other text markup). Span represents that. Div >> >>does >> >> > not. >> >> > > You are suggesting changing to div JUST IN CASE someone needs to >> >>set a >> >> > > width and/or truncate. Just-in-case code is not PAYG. >> >> > > >> >> > > My 2 cents, >> >> > > -Alex >> >> > > >> >> > > On 3/15/18, 2:17 PM, "[email protected] <mailto: >> >> > [email protected]> on behalf of Carlos Rovira" >> >> > > <[email protected] <mailto:[email protected]> on >>behalf >> >>of >> >> > [email protected] <mailto:[email protected]>> wrote: >> >> > > >> >> > >> And why not change IBeadView in Label for a view that uses a div >> >>when >> >> > >> needed? ;) >> >> > >> in the other cases, maintains the span >> >> > >> >> >> > >> 2018-03-15 21:05 GMT+01:00 Harbs <[email protected]>: >> >> > >> >> >> > >>> I’m still not understanding you. What about Label says it’s a >>span >> >> > >>> instead >> >> > >>> of a div? Why is making it a div not PAYG? Span simply limits >> >>Label >> >> > >>> unnecessarily. I think span is not the appropriate element for >> >>Label. >> >> > >>> >> >> > >>>> A Label without width should grow to the size of its text. >>Isn't >> >> that >> >> > >>> how >> >> > >>>> it worked in Flex? >> >> > >>> >> >> > >>> Sure, and a div does that too. In Flex, a label could have a >>width >> >> too >> >> > >>> if >> >> > >>> specified. Spans cannot. >> >> > >>> >> >> > >>> Harbs >> >> > >>> >> >> > >>>> On Mar 15, 2018, at 9:56 PM, Alex Harui >> >><[email protected]> >> >> > >>> wrote: >> >> > >>>> >> >> > >>>> Have the truncation bead swap the span for div. Again, some >> >> component >> >> > >>> has >> >> > >>>> to put a span in the DOM. Label has been just fine for that >>so >> >>far. >> >> > >>>> >> >> > >>>> A Label without width should grow to the size of its text. >>Isn't >> >> that >> >> > >>> how >> >> > >>>> it worked in Flex? >> >> > >>>> >> >> > >>>> Maybe Label should be refactored to not have width/height >> >>properties >> >> > >>> but >> >> > >>>> IMO we have bigger fish to fry. >> >> > >>>> >> >> > >>>> I'm sure Div is used to wrap text in some frameworks, but >>those >> >> > >>> frameworks >> >> > >>>> probably aren't PAYG. In Alina's app, Label has a usage >>score of >> >> 69, >> >> > >>> but >> >> > >>>> Label.width has a score of only 5. We will have her use >> >> > >>>> LabelWithTruncation or add the truncation bead on the few >>labels >> >> that >> >> > >>> need >> >> > >>>> >> >> > >>>> My 2 cents, >> >> > >>>> -Alex >> >> > >>>> >> >> > >>>> On 3/15/18, 12:45 PM, "Harbs" <[email protected]> wrote: >> >> > >>>> >> >> > >>>>> But we already have a truncation bead. The only reason it >> >>doesn’t >> >> > >>> work >> >> > >>> in >> >> > >>>>> Label is because spans don’t have widths. >> >> > >>>>> >> >> > >>>>> Label is not something which should flow beyond its bounds — >> >>which >> >> is >> >> > >>>>> what span is useful for. I still don’t see why you think span >> >>makes >> >> > >>> more >> >> > >>>>> sense. >> >> > >>>>> >> >> > >>>>> FWIW, div *is* used as a container for text in various >> >>frameworks. >> >> > >>>>> >> >> > >>>>> My $0.02, >> >> > >>>>> Harbs >> >> > >>>>> >> >> > >>>>>> On Mar 15, 2018, at 9:42 PM, Alex Harui >> >><[email protected] >> >> > >> >> > >>>>>> wrote: >> >> > >>>>>> >> >> > >>>>>> AIUI, "inline" is the default display style for Span. >>"block" >> >>is >> >> > >>> the >> >> > >>>>>> default for Div. HTML flows inline by default for just >>plain >> >> text. >> >> > >>>>>> Basic >> >> > >>>>>> components try to give Flex-friendly names for HTMLElements. >> >>Some >> >> > >>>>>> component in Basic needs to inject a span into the DOM. In >>a >> >>few >> >> > >>>>>> seconds >> >> > >>>>>> of looking at a Google web page, I did not see any divs with >> >>text >> >> as >> >> > >>>>>> children. I saw spans. The Divs all contained other >>elements. >> >> > >>>>>> >> >> > >>>>>> I don't create HTML web pages for a living, but if I were to >> >> place a >> >> > >>>>>> label >> >> > >>>>>> over a control like I often see in Flex, I would not wrap >>the >> >> label >> >> > >>> in >> >> > >>> a >> >> > >>>>>> Div. The control would be div, so I would us a span or just >> >>plain >> >> > >>> text. >> >> > >>>>>> If I looked at the final DOM and saw Divs around every >>chunk of >> >> > >>> text, I >> >> > >>>>>> would think there is something heavy about the framework. >> >> > >>>>>> >> >> > >>>>>> Again, the goal is to encapsulate common patterns. Label >> >>combined >> >> > >>> with >> >> > >>>>>> some control is trying to replicate common patterns for >> >>labeling >> >> > >>>>>> controls. >> >> > >>>>>> You are trying to solve the common pattern of truncating >>text. >> >> That >> >> > >>>>>> will >> >> > >>>>>> require a Div. But truncation, or even width/height is not >> >> something >> >> > >>>>>> that >> >> > >>>>>> should be built into Label for PAYG reasons. IMO, you >>should >> >> > >>> create a >> >> > >>>>>> different component with a different name. >>TruncatingLabel, or >> >> > >>>>>> LabelWIthTruncation. >> >> > >>>>>> >> >> > >>>>>> My 2 cents, >> >> > >>>>>> -Alex >> >> > >>>>>> >> >> > >>>>>> On 3/15/18, 12:19 PM, "Harbs" <[email protected]> wrote: >> >> > >>>>>> >> >> > >>>>>>> Why is span lighter than div? >> >> > >>>>>>> >> >> > >>>>>>> Inline and nowrap is not specific to span. AFAICT, there’s >> >> nothing >> >> > >>>>>>> specific to snap which fits Label better than div. >> >> > >>>>>>> >> >> > >>>>>>> As far as components which “does” span: Besides Span in >>HTML, >> >>we >> >> > >>> have >> >> > >>>>>>> HTMLText in Basic which encapsulates span. >> >> > >>>>>>> >> >> > >>>>>>> I can revert the change I made to Label, but I have not >>seen >> >>any >> >> > >>>>>>> downside >> >> > >>>>>>> to using div for Label. It seems to me that there’s >>nothing in >> >> the >> >> > >>>>>>> Label >> >> > >>>>>>> promise which dictates using span. >> >> > >>>>>>> >> >> > >>>>>>> Harbs >> >> > >>>>>>> >> >> > >>>>>>>> On Mar 15, 2018, at 6:55 PM, Alex Harui >> >> <[email protected] >> >> > > >> >> > >>>>>>>> wrote: >> >> > >>>>>>>> >> >> > >>>>>>>> >> >> > >>>>>>>> >> >> > >>>>>>>> On 3/15/18, 5:34 AM, "Harbs" <[email protected]> >>wrote: >> >> > >>>>>>>> >> >> > >>>>>>>>> Gotcha. FWIW, we now have the HTML element component set >> >>which >> >> > >>> more >> >> > >>>>>>>>> thinly wraps HTML elements. >> >> > >>>>>>>>> >> >> > >>>>>>>>> I can’t think of any reason why Label would be better as >>a >> >>span >> >> > >>> than >> >> > >>>>>>>>> a >> >> > >>>>>>>>> div. It seems to me that simply using div instead of span >> >>would >> >> > >>> make >> >> > >>>>>>>>> Label more versatile. Am I missing something? >> >> > >>>>>>>> >> >> > >>>>>>>> You may be missing that the only real goal of Royale is to >> >> > >>> encapsulate >> >> > >>>>>>>> popular patterns. Pretty sure I've seen span used on many >> >>web >> >> > >>> pages, >> >> > >>>>>>>> so >> >> > >>>>>>>> we need a component that does what it does, including >> >>assuming >> >> > >>> inline >> >> > >>>>>>>> instead of block display. Similarly, what would you think >> >>of a >> >> > >>>>>>>> framework >> >> > >>>>>>>> that had a HelloWorld that used a div instead of a span? >>You >> >> > >>> might >> >> > >>>>>>>> think >> >> > >>>>>>>> it is too heavy. >> >> > >>>>>>>> >> >> > >>>>>>>> Per PAYG, if span doesn’t' support certain features, then >>you >> >> > >>> create >> >> > >>> a >> >> > >>>>>>>> LabelWithTruncation and have it use a Div if you need to. >> >> > >>>>>>>> >> >> > >>>>>>>> My 2 cents, >> >> > >>>>>>>> -Alex >> >> > >>>>>>>>> >> >> > >>>>>>>>>> On Mar 15, 2018, at 2:30 PM, Peter Ent >> >><[email protected] >> >> > >> >> > >>>>>>>>>> wrote: >> >> > >>>>>>>>>> >> >> > >>>>>>>>>> Label was one of the very first components and the idea >>was >> >> > >>>>>>>>>> minimalism. >> >> > >>>>>>>>>> Button was <button> and TextInput was <input >>type="text"> >> >>and >> >> > >>> Label >> >> > >>>>>>>>>> was >> >> > >>>>>>>>>> <span>. >> >> > >>>>>>>>>> >> >> > >>>>>>>>>> I'm wondering if we should have NativeButton, >>NativeLabel, >> >> etc. >> >> > >>>>>>>>>> which >> >> > >>>>>>>>>> would be these minimal elements and have others that are >> >><div> >> >> > >>>>>>>>>> wrapping. >> >> > >>>>>>>>>> For instance, ImageAndTextButton is a <button><img >> >> > >>>>>>>>>> src="goo.jpg">Label >> >> > >>>>>>>>>> Here</button> which works but the alignment is weird so >> >>maybe >> >> > >>>>>>>>>> ImageAndTextButton should be a <div><img>Text</div> >>which >> >>can >> >> be >> >> > >>>>>>>>>> aligned >> >> > >>>>>>>>>> and styled better. >> >> > >>>>>>>>>> >> >> > >>>>>>>>>> Anyway, that's the reason: simplicity. >> >> > >>>>>>>>>> >> >> > >>>>>>>>>> ‹peter >> >> > >>>>>>>>>> >> >> > >>>>>>>>>> On 3/15/18, 7:51 AM, "Harbs" <[email protected]> >> wrote: >> >> > >>>>>>>>>> >> >> > >>>>>>>>>>> Is there a reason that the element type of Basic Label >>is >> >> span? >> >> > >>>>>>>>>>> >> >> > >>>>>>>>>>> I tried adding a łTextOverflow˛ bead to a Label and it >> >>does >> >> not >> >> > >>>>>>>>>>> seem >> >> > >>>>>>>>>>> to >> >> > >>>>>>>>>>> work because spans donąt really have a working width. >> >> Switching >> >> > >>> the >> >> > >>>>>>>>>>> element type to div seems to make it work. >> >> > >>>>>>>>>>> >> >> > >>>>>>>>>>> Harbs >> >> > >>>>>>>>>> >> >> > >>>>>>>>> >> >> > >>>>>>>> >> >> > >>>>>>> >> >> > >>>>>> >> >> > >>>>> >> >> > >>>> >> >> > >>> >> >> > >>> >> >> > >> >> >> > >> >> >> > >> -- >> >> > >> Carlos Rovira >> >> > >> https://na01.safelinks.protection.outlook.com/?url= >> >> > http%3A%2F%2Fabout.me%2 >> >><https://na01.safelinks.protection.outlook.com/?url= >> https%3A%2F%2Fna01.sa >> >>>>felinks&data=02%7C01%7Caharui%40adobe.com%7C3664f690b3da4e4aefab08d58ac >>>>c >> 7 >> >>bf0%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0% >> 7C636567533215453966&sdata >> >>=exoH%2BPKzuyM%2BykE65GQeHk%2BjPAZSsUFjsKwx3BRvjsk%3D&reserved=0. >> >> > protection.outlook.com/?url=http%3A%2F%2Fabout.me%2> >> >> > >> Fcarlosrovira&data=02%7C01%7Caharui%40adobe.com >> >>>><https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2F40adob >>>>e >> . >> >>>>com%2F&data=02%7C01%7Caharui%40adobe.com%7C3664f690b3da4e4aefab08d58acc >> 7b >> >>f0%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0% >> 7C636567533215453966&sdata= >> >>ewhtNMkSjHssIseUSvxq9etF%2FtrcA%2Buqk3sZlJkjHuc%3D&reserved=0 >> >> >% >> >> > 7Cae539925833d453aaa2e08d5 >> >> > >> 8aba29c4%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0% >> >> > 7C636567454874235113&s >> >> > >> data=Xvu%2BZ7186rOBNg5j0KrJsGPi5O4YnwiY1WVtz2xsNvc%3D&reserved=0 >> >> > >> >> > >> >> >> >>
