flex-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alex Harui <aha...@adobe.com>
Subject Re: Are we "nerd" driven or "business realistic" driven?
Date Thu, 27 Jun 2013 21:12:17 GMT
Hi,

Thanks for writing and providing more detail.  I have the following
questions and comments.

1) You recommended we talk to banks and other businesses.  I assume you
have.  Do you have any data you can share with us?  I've been pondering
how to get reliable data on what, if anything, folks are looking for.  In
December 2011, Adobe invited several major banks and other major Flex
customers to a summit.  In my conversations with the attendees, the desire
to have a migration path for their Flash-based Flex apps to HTML/JS/CSS
was my main takeaway, and that is what I've been working on with the
FlexJS prototypes.  Of course, it's been 18 months since then so things
could have changed, so if you have more recent data to share, it would
save the community some time.  Your list was much for Flash-based than I
would have expected, but maybe these banks are no longer in such a hurry
to migrate.

2) The list of changes you would like to see to current Flex matches up
pretty much with what we've seen and heard from the community.  Yes, we
all wish more could have been done regarding these topics in the past 18
months, but to be 'realistic', the number of developer hours spent on the
Flex SDK per-day has dropped significantly since Adobe donated Flex to
Apache.  It would be nice to see more folks get paid by their employers to
work full time on Apache Flex, but it hasn't happened.  And consider that,
in the area of performance, several Adobe engineers spent a lot of time
trying to find ways to improve performance and there aren't any global
hotspots left. Flex suffers from being "warm everywhere".  One of the
reasons the AS-side of the FlexJS framework uses small re-usable pieces is
in an attempt to take advantage of the JIT and run certain chunks of code
more often. And keep in mind that the current SDK is heavily tied to Flash
and its performance characteristics are outside the control of Apache
Flex.  So, we can all wish for more, but at this point, I think you have
to lower your expectations of how the current SDK can be made faster.
I've given up on anything other than the full rewrite I'm doing with
FlexJS.

3) It is certainly possible that FlexJS or any of the other variants of
Flex on JS may never be successful, but for sure, FlexJS does not have the
layers you claim it does. It just compiles MXML and AS to JS.  When I run
the examples, the JS versions actually start up faster than the SWF
versions.  And, when I get more databinding working, I'm going to start a
discussion about cutting releases of this stuff.  Yes, the first releases
will be 'alpha' quality, but it will get better faster by getting more
folks to bang on it.  Meanwhile, I have an internal "customer" that is
trying to migrate a real-world application on to FlexJS so I will be in
the loop on how this framework scales.  And anybody else is welcome to try
to port their applications as well.

4) Many of us in the Apache Flex community are doing a lot of work that
isn't sexy.  If you don't believe me, you try getting a mustella test
engine to work with Jenkins or fix dozens of date-localization bugs.  I am
doing what I think is on the top of the community's list that provides the
biggest bang for my time.  Continuing to look for small increments in
performance isn't going to have the pay-off of trying to make your apps
run without Flash.  But if you have data to the contrary, please share.

-Alex

On 6/27/13 11:42 AM, "Sebastian Zarzycki" <sebastian.zarzycki@gmail.com>
wrote:

>Hello,
>
>Sebastian Zarzycki here, author of that unfortunate post on
>spicefactory.org forums. First of all, apologies, if this will not hit
>the right list or will land in a wrong spot. Last time I've used mailing
>lists was around '95 and it's definitely not a user-friendly experience
>to hop in and respond quickly to a given post. But here goes.
>
>First of all, as mentioned already on spicefactory.org forums, my
>opinions are my own. I don't feel it was perfectly ok to bring up my
>response to this list, taking it out of the context and without notifying
>me. Might be oversensitive here, but I would probably play it
>differently. But that's a minor issue really.
>Second, I've started my Flex experience, when Flex 2 was in beta, and
>Flex 1.5 was paid server. This was long time ago (around 2005 I think).
>I've seen all the changes that happened. I've contributed a lot to Flex
>during those days, fixed lots of bugs. I even might still have somewhere
>my "Making Adobe Flex Better" shirt created and mailed out to Jira
>contributors back then. If you happen to have it, you will find out my
>name there. This is obviously not a bragging contest, but to make you
>realize, I'm here from (almost) the very beginning.
>
>I'd like to expand on that initial post, as I'm sure it feels quite
>negative and aggressive in that form. Wasn't my intention to undermine
>the work that was done on Apache Flex during that time and to discourage
>great people that devote their spare time to improve on things. My hats
>off to you. It's also easy to critic while staying aside and not doing
>anything to help - guilty as charged. But, you can tell that there is a
>lot of frustration in Flex community and this was probably a way of
>getting it out of my chest. I'm pretty sure, opinions on that will vary,
>but this is all right. Then again, I'm quite sure that there are lot of
>people out there who share my views. Whatever you do, doesn't change the
>flex market condition and that's a fact. I observe this list very often
>from the very beginning so I'm quite familiar with all the topics brought
>to life and the way they were tackled. So putting this everything again
>in a more consistent form:
>
>- I still stand by  "nothing anything (really) good with Flex".  Yes,
>this is harsh. Yes, it's unfair and significantly exaggerated. No, it's
>not really Apache fault. Please bear in mind that this is a bit different
>than "nothing good for Flex". It took a lot of time, but there's finally
>a good looking website with flashy statements and encouraging content (I
>always felt that Adobe was too shy with promoting Flex). There was this
>migration and some bugs are fixed - that's fine. But when you consider
>that so much time has passed and, in fact, so little done, I fear that
>the window frame we could actually turn the tides over is long past us.
>
>- It's not the problem of Apache or any other community. I still stand by
>"meritocracy" doesn't work. The reality is that certain pieces of work,
>especially in code, have to managed by one person. The person that has
>the general vision and when things go bad, can smash the fist on the
>table and say "No! It will be the way I want it to be!". That person can
>set up a plan, a roadmap and then find people who are going to fulfill
>that roadmap. This is (over-generalizing) what Adobe was doing and it was
>good. In Apache, we have "meritocracy" which basically means "work on
>whatever you want, whenever you want". This might be ok for smaller
>projects, but Flex is no small project. What's more, it's an idea and a
>global platform people invested a lot in. This sort of responsibility
>cannot be handled by "community". Community tends to come and go, flow
>by. Here you just need a solid vision and a solid development and
>marketing force. Without clear leadership and commitment, I simply don't
>see it. You might disagree, but then again, there's little in download
>section to prove it.
>
>- In general, I think that all Flex developers, despite all the internal
>framework mess and bad efficiency, love working with Flex. It's a
>brilliant technology and thanks to Flash Player / AIR, a dream come true.
>I've learned a lot how to be a good developer with Flex. It's natural
>that Flex developers are now disheartened and bitter. We've invested time
>to learn this and it's no longer paying off. But above everything else, I
>think that everyone feel deep inside, that it's not the business, but
>it's just sheer loss of such a great development tool to be constantly
>bashed, played down, laughed at and thrown away. We all know how cool it
>is. The world seem to never believe us. Isn't it this way? I'm sure most
>of you had a "so I've heard that HTML5 is great, convince me to Flex"
>situation for your contracts or in your companies. We fight, we struggle,
>and then all this energy is in vain now. This is extremely frustrating.
>
>- Internal bureaucracy is a fact. I don't want to go into a debate,
>whether it can be improved or whether it's necessary. I don't care. I
>just look from distance and judge the results. The results are far from
>satisfactory.
>
>- The way Apache's doing things feels very archaic to me sometimes.
>Mailing lists? Really? For new people, some might even not know what to
>do with bunch of emails presented on flex.org. Yes, there's a short note
>at the bottom, in a different section. Who will read that? Then we have
>links to apache and markmail. Case study - Markmail for users mailing
>lists doesn't work. Then we have "old" and "new" for Apache. Again,
>really? The result? During whole 2012 that list was almost barren. I see
>some more posts nowadays, but think of how many people you're (we are?)
>not reaching out to.
>
>- The GIT vs SVN argument was just something that popped out from my
>mind. Again, it's unfair to judge by this and it's exaggerated. It was
>just brought as example of how long it takes to do things and whether
>it's actually important. So much energy was spent on this, people
>convinced that GIT will be the next holy grail, and then after the move,
>it just brought a brand new chunk of frustration. This leads to the final
>-
>
>- "nerd" vs "business realistic". Again, apologies, if someone feels
>offended by the word "nerd". But the general impression I get from this
>list, is that instead of hardening and improving on Flex, this community
>has broken down into small groups of university science projects, that
>will take years before they get into "beta" stage and probably will never
>go stable, because market conditions will change. This is a good thing
>and a bad thing. The good thing is that, thanks to people who like to go
>beyond and tinker, we might get sometimes great things. The bad thing is
>that I believe that Flex don't need a revolution today and even if, we
>don't have resources to deliver such a revolution. All these Flex to JS
>projects I feel are typical "monkey arm" syndome. It looks great on paper
>and it's cool to work on this, but in reality, it will simply not work.
>We've seen this before, so many times - Echo2, GWT, etc. If you translate
>one layer (MXML) to another layer (AS) to another layer (ABC or whatnot)
>to another layer (JS), it's just too much. And you will never be able to
>compete with clean, direct JS frameworks as things change in that world
>so fast, that Flex compiler wouldn't be humanly able to keep up. Instead,
>I was surprised to learn that actual Flex problems weren't tackled. I
>might have inkling as to why it is so. Simply, because it's a lot of
>hard, tedious, not-sexy work and people in "meritocracy" have less
>incentive to commit to not-sexy work. Improving style handling feels
>pretty lame compared to ZOMG-FLEX2JS-COMPILER!!!!111one. Sad bad true.
>This is how human beings operate. Sometimes you have to be forced to do
>this ugly, tiring stuff to make a great product. Ask Apple.
>What Flex needs in my humble opinion?
>
>- improve on skinning/styling approach. It's a known fact, that Flex 4 is
>generally slower than Flex 3. That's no good.
>- get rid of mx totally and fill in missing spark parts
>- clean and easy to access 3rd party component market / shareplace
>- figure out the global chokepoints and improve on speed - either by
>changes in the framework, or by putting a pressure on Adobe to refresh
>Flash Player. Flex 4 performance is currently a laughing stock. It wasn't
>few years ago, but now it is, especially when compare to fast Javascript
>VM's. It's bizarre trying to explain your client that you have to
>virtualize lists and that itemrenderers are slow, even if you go knee
>deep in pure as approach. Yes, everything can be optimized, but it's
>matter of time and price. If you release a product, it has to deliver on
>what's promised. People don't have time to optimize on their own and
>learn the product. If you don't deliver, people will simply go away. And
>away then went.
>- arm Flex beautifully for mobile. I was a huge "nay" for Flex mobile in
>the beginning, but more and more I see point of it. Obviously, there's a
>lot here that depends on AIR, iOS/Android compiler and Flash player. But
>it doesn't mean that Flex cannot go forward with this.
>Where does that leave us?
>
>Again, I feel really bad that this landed on this list in such a way. I
>don't mean to insult or offend or discourage people who are working here
>on their projects. I respect your skills and your time which are often
>far far exceeding what I could even think is possible. But this is the
>"outside world" view, the "business realistic" view. Ask banks, who
>previously invested heavily in Flex. Ask software houses, interactive
>agencies. Ask Flex developers who mostly migrated towards greener
>pastures. Ask job-sites to find out that the amount of Flex jobs is very
>small. Apache Flex had a great challenge ahead and, I think, it did not
>deliver. Sure, it's easy to debunk saying "Apache is just people, it's
>you and me". But maybe this is the main problem. Food for thought.
>
>Best regards,
>Sebastian Zarzycki
>Feerie Software
>> Hi there,
>>
>> just wonder what you think about this forum post of the SpiceFactory
>>forum [1]:
>>
>> "With all due respect, I don't believe Apache did anything good with
>>Flex. They struggle with a lot of internal bureaucracy and I believe
>>that without clear leadership, this "meritocracy" thing just doesn't
>>work. For the past year, Apache Flex community debated whether they
>>should move to Git and right now when they did, they figured out that
>>most people don't know how to use it, etc. etc. In other words, to much
>>"nerd" driven, not enough "business realistic" driven."
>>
>> [1] http://www.spicefactory.org/forum/viewtopic.php?t=3803
>>
>> -- Sebastian (PPMC) Interaction Designer
>>
>> Looking for a Login Example with Apache Flex? Please check out this
>>code: http://code.google.com/p/masuland/wiki/LoginExample


Mime
View raw message