mlindgren.ca - Latest Comments

Re: Configuring DPI in Lubuntu/LXDE

Adam — Sun, 23 Feb 2020 07:14:13 -0000

ThanX man, it works like a charm.

Re: Configuring DPI in Lubuntu/LXDE

Mizumaky — Sat, 13 Oct 2018 04:57:59 -0000

Thank you! Helped a lot!

Re: Fitbit Ionic SDK impressions

Mitch Lindgren — Wed, 11 Jul 2018 20:59:16 -0000

Good to know. Thanks!

Re: Fitbit Ionic SDK impressions

Edwin Finch — Wed, 11 Jul 2018 19:30:49 -0000

Darn Americans ruining the way the English language was meant to be used, but that's a discussion for another day.

As for the SDK: the FitbitOS SDK has made massive strides since we last spoke here seven months ago - its speed, functionality and scope have all gotten better in their own ways, to the point where I can now write fully functional watchfaces on the SDK and they load in under ten seconds ;)

It's worth checking out again if you want to tinker with it, the community has also grown, so you'll find friendly help if you get stuck anywhere.

Although I still miss Pebble's C SDK dearly, it's nice to see that Fitbit is making the best of their situation (that they put themselves in) and I have confidence they'll continue to make worthy improvements with time.

Re: Fitbit Ionic SDK impressions

Mitch Lindgren — Tue, 10 Jul 2018 22:03:22 -0000

I was just re-reading this while reflecting on the current state of the Ionic, and I realized that I never answered your last question. I tend to use American spellings out of habit, because that is the convention in the vast majority of programming languages and frameworks. Plus, since I live in the US now, my operating system is set to the en-US locale, so it marks Canadian spellings as incorrect. :)

Back to the Ionic, I haven't really looked at the SDK in any depth since I originally wrote this post. Have you done any more work with it, and if so, has it improved?

Re: How to manually send a Pingback

Mitch Lindgren — Sat, 30 Jun 2018 17:08:30 -0000

Hi Joanna,
Unfortunately I haven’t tried this in a long time, so I’m not sure exactly what the problem is. As you stated, the source page must contain a link to the target page, but other than that I’m not aware of any specific requirements. Unfortunately I haven’t actually tried this myself since I originally posted it, and there’s a good chance something has changed in the past few years. You might be able to get a better answer on stackoverflow.com. Good luck!

Re: How to manually send a Pingback

Joanna Siwiec — Sat, 30 Jun 2018 16:11:31 -0000

Maybe the source page must contain specific content. Certainly it must contain a link to which pingback sends. But what else does it have to pass a positive verification?

Re: How to manually send a Pingback

Joanna Siwiec — Sat, 30 Jun 2018 16:07:22 -0000

It not work for me. I have response:

<methodresponse>
<fault>
<value>
<struct>
<member>
<name>faultCode</name>
<value><int>0</int></value>
</member>
<member>
<name>faultString</name>
<value><string>Invalid discovery target</string></value>
</member>
</struct>
</value>
</fault>
</methodresponse>
What is wrong?

Maby problem is in source url. Becouse when I try to post pingback from one WP blog to another it is work excelent.

Re: Fitbit Ionic SDK impressions

Edwin Finch — Sat, 09 Dec 2017 11:06:22 -0000

Thanks for your reply Mitch!

I was quite excited when I found your long spiel, haha. Aside from short-written thoughts on the Fitbit discord, I actually had not heard many people's opinions on the SDK. Let alone an opinion from a senior developer with lots of experience under his belt, who's very well spoken, so it was quite a treat to read what you had to say!

That's a very good point you made about my learning of C, demonstrating your earlier point of programming languages not being the main influencing factor in a developer's interest of an SDK. I truly was so passionate about Pebble that I did not care about all of the learning curves, I just wanted to write some apps for my wrist. I thought many other folks in the Pebble community felt the same as well, after all, their appstore was quite comprehensive and the selection of apps was impressive. I wonder if Pebble thought the opposite, did they think that they did not have enough developers, or that there wasn't enough of a selection of apps...? My question now is why Fitbit thought it'd be a good idea to change gears from a smooth C SDK to a clunky JS SDK, when it seemed the C SDK had already proven itself.

As for JIT/AOT compilation, I had to look into these terms as I did not know them, thank you for teaching me something new :). I therefore don't know much about it, though it seems that your explanation is logical, that the small performance gains from JIT would not be worth abandoning AOT in Studio.

It is quite funny how even first party apps are suffering from speed issues. With that, I hope speed improvements are a priority for them. It'd make the experience a lot smoother on both the developer's and user's ends.

In the end, if speed improvements are made, I'll be back in the game right away, despite all of the frustrations with JavaScript. Developing for wearables, along with what seems to be an already strong community of developers (although small), makes for a very fun experience. I hope eventually it'll be more encouraging for you too :)

(Off topic question, I notice your site's TLD is .ca, but you spelled favourite wrong in your article? ;)

Re: Fitbit Ionic SDK impressions

Mitch Lindgren — Sat, 09 Dec 2017 00:32:10 -0000

Hey Edwin,
Thanks for the great comment. It's very gratifying to me that someone actually read this long spiel.

First off, I just want to say that it's quite impressive to me that you were able to teach yourself C in the course of developing Pebble apps. C is not an easy language to learn, and I didn't really grasp its nuances until I was in college working towards a computer science degree - and I already had a decent amount of programming experience in other languages at that point. Teaching yourself C is no small feat, but I think the fact that you did so because you wanted to develop Pebble apps reinforces my point that the ease of use of the programming language is not the primary factor in determining whether or not a platform will be successful. In other words, if the device itself is good and has some commercial success, and the development platform *works*, developers will come to it, even if it's not that easy to develop for.

I didn't realize that apps are not compiled to bytecode until the first time they are run on the watch. I don't think your question about why it doesn't happen in Fitbit Studio is ignorant at all. The only thing I can think of is that perhaps the runtime is doing JIT compilation as the app runs and then just caching the bytecode once all of the code has been JIT'd. JIT compilation does have some advantages over ahead-of-time compilation, as the compiler can use profiling to make optimizations that aren't possible with AOT compilation. But I'm not sure that's actually happening, and even if it is, it seems like a bad decision by Fitbit. On a low-powered CPU like the Ionic's, it seems to me that the minor performance gains from JIT compilation vs AOT would not be worth the massive performance penalty the first time the app is run. To be honest, I'm surprised the Ionic itself even has the capability to compile raw JavaScript to bytecode. It would make more sense to me to have it *only* understand bytecode.

(In fairness, I've never worked on an app platform either, and JavaScript is not my specialty. As the original post probably made clear, I try to avoid it as much as possible. But I can at least speak as a professional software developer with several years of industry experience, and from that perspective, nothing about having the watch compile the code makes any sense to me.)

At any rate, it's ironic that your app got rejected for taking too long to load, because the first-party Strava app has the same issue.

Obviously, I share your disappointment in the state of the SDK so far. I have a couple ideas for apps I'd like to implement, but frankly, I have very little motivation to spend any time working on these apps with the SDK in its current state. That said, I have found some encouragement in the latest firmware update which made the app gallery available. There are already a bunch of cool apps and watchfaces available, which gives me more hope for the long-term health of the platform. I really like the Ionic, and I can see that the Fitbit devs have put a lot of hard work and passion into it, so I want it to succeed. Hopefully the problems we've discussed will prove to just be growing pains.

Re: Fitbit Ionic SDK impressions

Edwin Finch — Fri, 08 Dec 2017 23:47:50 -0000

Excellent read, and all very well said.

I might have something to add, as I've spent a lot of quality time with the Pebble SDK (since its 2.0 beta). I have also spent around a month with the Fitbit SDK, developing 3 watchfaces for the Ionic.

Your article has outlined quite perfectly the situation Fitbit has created by choosing JavaScript as their language for app development. I literally taught myself programming by starting on the Pebble - forcing myself through, what at the time, were difficulties of C. Pointers as a complete programming noob, anyone? I now realize I took *a lot* of what Pebble's C SDK offered for granted.

Back in the Pebble days, I wrote a collection of watchfaces along with my own payment solution (Pebble did not have their own payment solution either). Users could purchase our watchfaces for $1.25 each, or $5.99 for all of them. These were my early days of programming, keep in mind.

I attempted to build the same payment solution on Fitbit's platform, along with 3 watchfaces for the Ionic, in hopes of releasing these on the gallery. I noticed very early on in my development journey, though hoped that it would not be as bad as it turned out to be: apps developed on Fitbit's SDK are slow. And I'm talking slooooow.

When you "compile" your app in Fitbit Studio, all of the code you write for the watch is put into a single file, index.js. This massive file, when first run on Ionic, is compiled into bytecode on the watch. That bytecode is cached, making subsequent loads of the app around twice as fast, in my experience.

My problem is exactly that, though. Subsequent loads. That first load is absolutely killer, especially when Fitbit kills apps for being "unresponsive" if they take over 10 seconds to load. I do not understand why bytecode compilation has to happen on the watch. Why does bytecode compilation not happen during compilation within Studio? I of course have no experience in building an app platform or SDK, so I may be questioning out of ignorance, though I figure it'd probably be easy to do so and would make the whole experience a lot smoother.

To cut a long story short: I spent a long time battling these speed issues with Fitbit's SDK. I'm not the completely inexperienced developer that I was when I wrote Pebble watchfaces (which all loaded in under half a second, if not instantly). I have more experience under my belt. I'm still young and have *a lot* to learn, and comparatively to many developers I am still quite inexperienced, but I can assure you that the code I am writing these days is comparatively more efficient and cleaner than before. Yet, I unfortunately cannot get my app's first load time to be under 10 seconds.

Well, that's not completely true, I did figure out a way. I load my compiled index.js through the app's resources folder using `require()` when initializing. Since the SDK performs tree shaking, I can write my app as I wish. Then when I want to give it a "speed boost" (aka get the app to load in under 10 seconds); I take my app's index.js, insert it in my resources folder, and make that `require()` line the only line in my app/index.js in Studio. Boom, no bytecode compilation & caching on the watch. Boom, 9 second load time, woohoo!

Boom, I was unable to get all 3 of those watchfaces approved for distribution. The reason for all 3? You guessed it: lack of speed. It's one of Fitbit's gallery requirements: all apps must load in under 5 seconds, after bytecode compilation & caching has taken place. Unfortunately my workaround means 9 seconds across the board, even if my app is never forcibly killed for being "unresponsive".

The Fitbit team is quite aware of the issues I've been having (I wonder if I have been too vocal with them sometimes). They tell me that they're working on speed improvements, and I'm happy to help them do so. Regardless, this first experience has been quite frustrating and discouraging, especially as a developer who loves developing for wearables.

I'd also like to bring up your comments about Fitbit's supposed takeaways from Pebble. What you talked about reminds me of a question I often ask myself. Aside from the employees, did Fitbit actually use anything from Pebble? It seems as though a lot of it has gone to waste. The speed, functionality, and more, have all seemed to disappear without a trace. I love Pebble, and it breaks my heart to see Fitbit's SDK struggling like this. I know it's only the first version, and that Pebble had its issues as well, though so far it has been painful to watch.

On a positive note: I look forward to seeing where the SDK goes and hope that my month of work on 3 watchfaces and a payment system does not completely go to waste. I'd love to see the community grow, the SDK flourish and the company prosper :)

P.S. Here's an interesting discovery I made with my speed struggles, that you may find fascinating.

https://www.youtube.com/wat...

Re: Configuring DPI in Lubuntu/LXDE

Aaron N. Brock — Fri, 13 Oct 2017 17:24:59 -0000

This only scaled up my text on my display, all the images are still tiny!

Re: Configuring DPI in Lubuntu/LXDE

Brian Nguyen — Sat, 24 Jun 2017 17:55:31 -0000

This works great, thank you so much.

Re: Configuring DPI in Lubuntu/LXDE

dusan — Fri, 03 Jun 2016 15:37:43 -0000

for lubuntu 16.04 live usb - sudo leafpad /usr/share/lightdm/lightdm.conf.d/50-xserver-command.conf

Re: Configuring DPI in Lubuntu/LXDE

Mitch Lindgren — Fri, 27 Nov 2015 18:10:48 -0000

If the application doesn't already respect the X DPI settings, I'm afraid I don't know how to change it. Gimp may have its own DPI controls but I'm not familiar with it myself.

Re: Configuring DPI in Lubuntu/LXDE

Alexandre Ruiz — Fri, 27 Nov 2015 17:58:02 -0000

Thanks for that. Works like a charm.

Do you know any way to fix the icons size from other applications? Like in the toolbox of Gimp... Still small...

Thanks again.

Re: Observations from Three Years in America

ThatGuy — Fri, 11 Sep 2015 16:54:39 -0000

Thanks for the DPI Lubuntu fix, life saver and I couldn't find it anywhere! I was about to move on when I realized you're a Canadian immigrant to the US and I read your article. My Dad was the same about 40 years ago, during the 70s, which I've heard bears many similarities to current time. I have lots of family in Canada, but sadly I haven't visited in quite some time. I've thought about going the other way, migrating to Canada. But unfortunately I'm not a big fan of the cold and I'm too heavily invested in my "American" self identity. I've served in the US military and since we're talking politics, my personal belief is that's what makes you a "citizen". Temporarily or permanently giving up the privileged life that came with the luck of being born into one of the best countries in the world to "serve" this country. It doesn't have to be the military, could be non-military federal service, but something that gives back or earns the right to call oneself a citizen beyond taxes or money. I've also worked in IT for 20 years, so when you talk about H1B, I don't think about the five Canadians, who look, talk, act, and as you said are "culturally" like me. Instead I think of the Indians, Filipinos, and Chinese that have mobbed the tech centers of the west coast completely changing the demographics and culture of those areas and the companies. I don't care who you are that is a difficult pill to swallow, seeing such a drastic change in your environment in such a short period of time. If I took a million religious Texans and transplanted them to a few neighborhoods in Vancouver, Shanghai, Manila, or Mumbai what do you think the reaction would be? Now add to that a multitude of other long standing American homegrown domestic societal issues...minimal public education, lack of critical thinking, lack of civic involvement, corporate money politics, "bread and circus" (welfare, fantasy sports, reality tv, drugs), lack of corporate trust, lack of government trust, etc...It becomes difficult not to believe H1B and immigration is a scheme by corporations and their bought and paid for politicians to increase profits, guarantee future votes, and undermine the existing middle class. I don't buy the "only in the USA" argument, the same people we are bringing here could be contributing the exact same way at Apple, Microsoft, and Intel in India, China, and the Philippines. In a constantly connected world, smart talented people can't create things regardless of their physical location? Its the same stupidity that keeps people commuting 2 hours a day when they could be doing the same work from home. I suspect its all related to our bass-ackward income tax system. We refuse to change the way we generate tax revenues, so we need warm bodies from other countries to come here and pay for the welfare, healthcare, and retirement of our lazy natural born "citizens". So welcome H1Bs the joke is on you, you are here to pay for our handouts! Maybe you can fix the giant mess left by "The Worst Generation" in American history...the Baby Boomers!

Re: Observations from Three Years in America

Laurie M Friskie — Mon, 07 Sep 2015 22:59:53 -0000

Canada's loss but I am happy that you are happy there.

Re: Configuring DPI in Lubuntu/LXDE

ASFP — Mon, 03 Aug 2015 04:03:43 -0000

Oh thanks a bunch!!! such an easy solution, finally! Good job

Re: Configuring DPI in Lubuntu/LXDE

Mitch Lindgren — Tue, 28 Jul 2015 17:31:37 -0000

Glad to be of help!

Re: Configuring DPI in Lubuntu/LXDE

Mitch Lindgren — Mon, 25 May 2015 16:51:16 -0000

You're welcome! I'm glad this post helped you out.

Re: Configuring DPI in Lubuntu/LXDE

Joel — Mon, 25 May 2015 15:40:32 -0000

Thank you sooo much!
I installed Kodibuntu and when booting in to Lubuntu everything was enormous, but by following your steps (except I changed the dpi to 96 since I wanted it smaller) everything now looks great!
It was quite hard to find the solution;everywhere I read about editing xorg.conf, which is not present in my installation. Once, again; Thanks! You saved me hours of fiddling around ;-)
Best regards
Joel

Re: Fun with the Microsoft Band SDK Preview

Jamie warren — Thu, 19 Mar 2015 05:55:07 -0000

Ah, not what i was hoping for as i need the raw PPG data! Thanks very much for your help Mitch, much appreciated.

Re: Fun with the Microsoft Band SDK Preview

Mitch Lindgren — Thu, 19 Mar 2015 04:08:08 -0000

Hey Jamie, I just double-checked and verified that what I said previously is correct: there is only one supported interval for heart rate, which is 1Hz.

Re: Fun with the Microsoft Band SDK Preview

Mitch Lindgren — Wed, 18 Mar 2015 13:05:29 -0000

Hi Jamie,
I'll have to double check this for you later, but if I recall correctly the only reporting interval available for heart rate is 1Hz.