PodSearch

Under the Radar

67: Dealing with Old Code

 

00:00:00   Welcome to Under the Radar, a show about independent iOS app development.

00:00:03   I'm Mark Orment.

00:00:04   >> And I'm David Smith.

00:00:06   Under the Radar is never longer than 30 minutes, so let's get started.

00:00:10   So today we are going to be talking about old code.

00:00:15   And the idea for today's discussion got started from some work I've been doing recently,

00:00:20   updating and fixing some issues in one of my apps.

00:00:23   It was actually the first app that I had that ever really was in any reasonable way a success.

00:00:28   It's an app called Audiobooks that lets you listen to public domain audiobook recordings.

00:00:34   And I launched it in spring of 2009.

00:00:39   So it's coming up on its eighth birthday today.

00:00:44   And it's kind of a wild thing to be going through and working -- I'm still actively

00:00:50   working on it for the last eight years.

00:00:52   I've had continuous updates, and it launched at a time when there was one screen size.

00:00:58   And that screen size was non-retina.

00:00:59   We were doing retain and release.

00:01:03   All of the buttons are super shiny.

00:01:05   They have that really round kind of capsule look to them.

00:01:10   And it even has some features that I look at now, and it took me a while to remember

00:01:14   why I added them.

00:01:16   For example, it has a feature where you can automatically resume playback of the audio

00:01:21   when the app starts up.

00:01:23   And that's because the app was created before multitasking was a thing, where when you close

00:01:28   the app, it would stop playing.

00:01:31   And so I made it so that as soon as you start it back up, you could resume to make it a

00:01:37   little bit easier for that.

00:01:38   But the app is so old that it has all of these code paths and these details.

00:01:45   There's sections of the code where it says, "Is multitasking supported?

00:01:49   Do this.

00:01:50   If not, do that."

00:01:51   And multitasking has been available since iOS 4.

00:01:54   >> Enoch: And if I'm doing my timeline correctly, it looks like you launched this when the current

00:01:58   version of iOS was still 2.0.

00:02:00   >>

00:02:23   The thing that's been making me think about, though, is I'm doing this process of doing

00:02:41   this big spring cleaning for this app.

00:02:44   And it creates all these weird questions, though, like how to deal with all this old

00:02:49   code, all this old stuff, and there's all this legacy weight that goes into this app.

00:02:54   And it's theoretically possible that there's a customer out there who has been using the

00:03:00   app since 2009 continuously.

00:03:04   And so there's all these interesting cases and details that I have to manage and deal

00:03:08   with because, for example, they may have had data in their database that is from the original

00:03:16   version, and I have a whole bunch of code in my app to translate old versions of, you

00:03:22   know, I think I've gone through three or four database formats or things.

00:03:25   Like originally I launched and the database was a plist, I think.

00:03:31   >> When you launched, core data was not yet on the phone.

00:03:34   You could do raw SQLite through the C API, but most people just had these weird little

00:03:39   hacks like plists or very, very basic SQLite functionality because it was so cumbersome.

00:03:46   >> Yeah, exactly.

00:03:48   It was a plist to start with, and then I transitioned to core data, and then core data has gone

00:03:53   through several migrations.

00:03:54   And so it's a lot of these weird, complicated issues that I feel like you don't -- when

00:04:01   I'm building things now, I don't really think about this kind of future-proofing and migration

00:04:04   and translation and the things that I may need to do if the app is sitting around.

00:04:09   I don't know if you think about this when you're doing overcast stuff, that like what's

00:04:13   going to happen if someone who's been running the old version for years and years and then

00:04:20   suddenly installs the new version?

00:04:21   It could have a totally different database scheme.

00:04:23   It could have a totally different file format.

00:04:26   All the MP3s could be in a different place.

00:04:29   Time is a weird thing with software, I think.

00:04:32   >> Yeah, I mean, I've avoided many of the -- I mean, first of all, my migrations for

00:04:37   the database are all fairly straightforward.

00:04:40   I do have a lot of that code that can support old versions.

00:04:43   However, in practice, I don't have a lot of people using old versions because from day

00:04:47   one I decided to use SSL certificate pinning on my connection to my API.

00:04:52   And this has caused -- it's a nice thing for security.

00:04:56   However, at various points I've tried pinning two various things.

00:05:00   First I pinned to the certificate itself, which only lasted until that SSL certificate

00:05:05   expired, which was about a year.

00:05:08   And then I realized, well, it's kind of inconvenient to break all the old versions of my app every

00:05:11   time I have to update my SSL certificate on my server.

00:05:15   So I tried pinning to the intermediate certificate through my SSL issuer, figuring, well, that

00:05:19   won't change for a long time until it expires in like five years.

00:05:23   And then all the encryption standards got upgraded to, I think, SHA-256 or somehow they

00:05:29   changed the signature algorithms to be more secure on all certificates.

00:05:32   And so that broke.

00:05:33   And so I had that force an upgrade.

00:05:35   And then on my most recent one, just a few weeks ago, I updated my certificate.

00:05:41   And I'm like, oh, I'll be fine.

00:05:42   I'm pinned to the new SHA-256 intermediate certificate.

00:05:45   I can update and it'll be fine.

00:05:47   But then the new certificate I got just was signed with a different intermediate key.

00:05:52   Like for whatever reason, like, OK, that's not good.

00:05:57   So I had to forcibly break all previous clients who connect.

00:06:01   And I gave like a month notice and I let it go for a while.

00:06:05   But then like the actual expiration date on the old certificate was actually as we record

00:06:11   yesterday.

00:06:12   And so I was forced to break all compatibility with all old versions yesterday, basically.

00:06:21   So I don't really have that problem as much.

00:06:25   And this kind of gives me a lot of advantages.

00:06:27   I mean, one advantage is that I don't have to really worry about supporting old versions

00:06:31   of the OS very much.

00:06:32   It's like, well, even though it's inconvenient that I had to force all these people to update

00:06:36   the latest version, that also means that I literally have zero people using anything

00:06:41   older than it.

00:06:43   So it simplifies a lot of this stuff.

00:06:47   And I've found that usually people who are running very, very, very old versions of things,

00:06:53   they kind of expect things to break if they ever update them.

00:06:56   That's one of the reasons that they aren't updating them, because either they can't or

00:06:59   they won't.

00:07:00   And the reasons why they can't or won't are usually because either their hardware doesn't

00:07:03   support anything new, or in which case you're fine, because any new installation will be

00:07:07   on new hardware.

00:07:08   Or that they refuse to update for whatever reason.

00:07:12   Maybe they just don't like the new OS.

00:07:13   Maybe they think it'll run too slowly on their hardware.

00:07:15   Maybe they have a jailbreak they want to preserve.

00:07:16   Whatever the case is, they refuse to update.

00:07:19   In which case, I feel like if you are refusing to update to something that you could update

00:07:23   to, I kind of feel like you're taking matters into your own hands and accepting the responsibility

00:07:27   of things might be weird for you holding onto old stuff.

00:07:31   And one of the ways things might be weird is you might have some kind of failed migration

00:07:35   in the future when you do decide to upgrade, when you're upgrading way after everyone else

00:07:39   has.

00:07:40   So for the most part, the issue of migrating old data over time and everything is, I think,

00:07:47   not that big of a deal in practice.

00:07:49   You should do it as much as you can.

00:07:51   You should accommodate that, but not to the point where it's going to become a large burden

00:07:55   on you while you should be supporting new customers, basically.

00:07:59   >>

00:07:59   Yeah.

00:08:00   And I think it certainly is complicated, especially for apps that, like Overcast, it requires the

00:08:06   server for it to function, period.

00:08:09   And so it doesn't have the sense of an offline or a standalone kind of feeling.

00:08:15   Where this app, my audiobooks app, it still works.

00:08:19   I actually, out of curiosity yesterday, I exported version one out of my Git repository

00:08:26   and got it to run in the most recent version of Xcode, which surprisingly worked surprisingly

00:08:33   well given how much time has passed between those things.

00:08:37   >> That's kind of impressive.

00:08:38   >> I only had to fix a few bugs and set the root view controller on the window and a few

00:08:43   things that otherwise it just wouldn't launch.

00:08:45   But overall, it ran.

00:08:47   And the app still works because it was built at a time when it needed to be standalone

00:08:55   because network connections were super flaky and everyone was on edge or 3G at the time.

00:09:01   And it was a very different paradigm.

00:09:03   So the app was very standalone.

00:09:05   And I think any app that has that kind of standalone-ness to it is where it really gets

00:09:09   complicated for these kinds of data formats and migration kind of stuff.

00:09:15   What I'm struggling with now as I go through this is the issue you run into is you end

00:09:20   up with this kind of Russian doll data format where I have code in the app that lets me

00:09:29   migrate -- every little migration or thing that I've done over time, I have code to manage

00:09:36   from one to the next to the next to the next.

00:09:38   And so there is a path that if I took something that was running that very first version,

00:09:44   I think there's -- maybe not that very first one, but definitely the core data versions.

00:09:49   Theoretically, if you had a device that was running that and then you reinstall the latest

00:09:53   version, it would go through six different migrations.

00:09:57   It would migrate this format change, then this format change, then this format change.

00:10:02   And it's a really tricky thing, though, for me as I'm going through this code.

00:10:07   I have no idea if this works still.

00:10:10   But it's still there in the code that if you were running database version 2, that

00:10:14   I have this code that says if you're 2, I can get you to 3, and then there's code from

00:10:17   like 3 to 4.

00:10:19   But that sounds horribly error-prone and unlikely to actually work in practice.

00:10:24   And so what I'm trying to go through now is should I just be deleting some of these old

00:10:28   code paths?

00:10:29   Should I just be going through and saying I'm supporting migration from whatever the

00:10:34   last major version to the current version and then cutting it off there?

00:10:39   Because it's a really -- theoretically, if I'm having this code in there that's doing

00:10:43   something, I should probably be able to -- I should probably have a way to reliably ensure

00:10:49   that it works.

00:10:51   And I don't know if that's actually a guarantee that I want to make.

00:10:55   >> Yeah, I mean, like, you know, on one hand, you could just do what you just did and check

00:10:59   out version 1.0 out of Git and build it and put some data in there and then check out

00:11:04   the current head version and run it and see if everything gets migrated over.

00:11:08   But again, I think as indie developers, there's so much stuff where like, well, it would be

00:11:14   nice if we had the time to do X, Y, or Z.

00:11:16   Or it would be nice if we had enough staff and resources to do X, Y, or Z.

00:11:20   But in reality, you have to prioritize.

00:11:22   And it is probably not worth a lot of effort to maintain these code paths that are going

00:11:30   to support somewhere between like 0 and 10 people ever.

00:11:35   You know, like it's not like a huge deal.

00:11:37   So basically, like the way I come down on it is, you know, what I said earlier, it's

00:11:41   like if it's going to be really easy to keep the support in, like if you have a couple

00:11:46   of instructions to migrate databases forward, like I don't see the value in deleting them

00:11:51   necessarily because, you know, that isn't that much code and it's not really hurting

00:11:55   anything to be there.

00:11:56   But I, you know, but if it becomes a burden, if for whatever reason that is causing you

00:12:01   to not be able to upgrade to something else or clean up something else or not be able

00:12:07   to change the app in a way that would benefit all current and new customers, then I think,

00:12:14   you know, then the cost is too great to having those and then you should get rid of them.

00:12:17   But I also don't think it's going to be worth your time to routinely check out version

00:12:23   1.0 and get some data and then check out the head.

00:12:27   To test these migrations that far back where you might literally be the only person who's

00:12:32   ever doing that, then it's probably not worth the time taken away from doing things

00:12:38   that benefit everybody.

00:12:39   Yeah, and I mean I think in a similar way I've been also working through some of these

00:12:43   thoughts about dealing with like just old device styles and types too.

00:12:49   Like for example, I still have all non-retina assets in the app.

00:12:55   Oh, geez.

00:12:56   Which is a funny thing because on the iPad, does iOS 10 have a non-retina iPad that it

00:13:03   still supports?

00:13:04   No, because even the iPad 3 it got dropped.

00:13:07   But it is, I think 9 still supported iPad 2 and 3 I think.

00:13:12   Yeah, and so it is funny when I have all these resources, especially if they're iPhone only,

00:13:17   like these are resources that can never be seen.

00:13:21   And so I'm going through and being like, "Should I clear this up?

00:13:24   It'll make my bundle a little bit smaller to just have 2x and 3x assets."

00:13:30   And I mean on the iPad it's especially interesting because for me and my use, something like

00:13:36   45% of the iPad users who use audiobooks are doing it on a device that doesn't support

00:13:42   iOS 10.

00:13:43   Wow.

00:13:44   Like mostly the iPad 2 and the iPad mini.

00:13:47   That's a lot.

00:13:49   Yeah, which makes up, I mean I imagine it's the nature of the, I mean the app's just been

00:13:53   around for a long time and that's active use too.

00:13:56   It's not just like total install.

00:13:58   Because I can imagine it's a lot of people who have an iPad that they just load audiobooks

00:14:02   on and listen to in their kitchen or on their bedside when they're going to bed or something

00:14:10   like that.

00:14:11   It seems like it's that kind of a device.

00:14:12   And so it's tricky when I have these devices that, I think right now I support iOS 8, 9,

00:14:20   and 10, or is my official support, and with this update I'll probably be going to 9 and

00:14:25   10.

00:14:26   But one thing that I've definitely run into a few times is dealing with these, is the

00:14:30   issue of a device that just can't get the latest version of the app, period.

00:14:35   And I think this is, with the iPads especially, this is going to be more notable than it is

00:14:41   for iPhones.

00:14:42   Because I just think iPhones don't stick around as long.

00:14:45   Like there's very few people using an iPhone 4 or an iPhone 3GS these days.

00:14:51   And so it's something that I worry about less.

00:14:54   Though I will say one little trick that I always do, I've started doing in all my apps

00:14:59   just like for, it's saved me so many times, is I always enable the iTunes file sharing

00:15:08   setting in most of my apps.

00:15:12   And it's mostly because it's like, I came into it most usefully for my recipe app, where

00:15:18   it has people store their recipes and the images and all these associated assets for

00:15:26   the recipe collection.

00:15:27   And it's nice to have a method that if someone emails me and says, I get this a lot, is I

00:15:32   have a first gen iPad, like my recipe book launched day one, or pretty soon after that,

00:15:38   of the original iPad.

00:15:39   And there are people who've been using it just like they take that iPad, they put it

00:15:42   in their kitchen, and they never changed it.

00:15:43   And they've been using it ever since.

00:15:46   And that version of the iPad is so unsupported by iOS versions, by versions of the app, that

00:15:54   it was really awkward to be like, well, there's not much that I can do to help you.

00:16:00   And so what I ended up saying is I enabled that iTunes file sharing.

00:16:03   And then at least what I can say to these people is, hook it up to iTunes, grab these

00:16:07   files, put them on your desktop, drag them back onto your new iPad, because usually that's

00:16:11   the situation they find themselves in.

00:16:13   It's like, I got a new iPad, yay, but what do I do?

00:16:15   And it's like they can then, at the very least, take those database files and move them back

00:16:19   and forth.

00:16:20   And then, like I was just saying, I have all the Russian doll migration scripting so that

00:16:25   it can take those first early versions of the app and make them move forward.

00:16:30   But if I hadn't enabled the iTunes file sharing API back then, they would just be stuck.

00:16:37   There'd be no way to get those files realistically out of the device.

00:16:40   They'd have to use some kind of iPhone Explorer specific tool.

00:16:45   Download Phone View.

00:16:46   Yeah, download Phone View and migrate, which is never going to happen.

00:16:49   But at least iTunes.

00:16:50   I think if step one of your troubleshooting is download Phone View, you're probably not

00:16:55   going to be successful with that.

00:16:58   So at least that is a little hint that I now just do it out of course, because I think

00:17:02   most people don't know it's there, and so it's unlikely to be dangerous.

00:17:07   Somehow people are going to be like, oh, what can I do with these files?

00:17:09   And start randomly deleting their databases.

00:17:12   But every now and then, it really comes in handy for somebody who's like, how do I move

00:17:17   my files from this phone to this phone?

00:17:20   Or it's even the common one that I get now with like, in pedometer, people are like,

00:17:24   I have all my steps on this device, and I don't want to do backup and restore, or it's

00:17:29   too old to do backup and restore because the versions are incompatible.

00:17:32   What can I do?

00:17:33   And it's like, well, go to iTunes.

00:17:35   It's awkward, but you just write out a big customer support guide that says click here,

00:17:41   click here, click here, and then it works well enough.

00:17:44   All right, we are sponsored this week by Linode.

00:17:47   Linode servers offer industry-leading performance with native SSD storage, Intel E5 Xeons, and

00:17:54   access to a 40-gigabit network.

00:17:56   They have nine data centers spread across the world, which give you the ability to serve

00:17:59   your customers quickly.

00:18:01   And Linode's API allows you to easily automate tasks or develop custom applications in the

00:18:05   cloud with super simple scaling.

00:18:08   This allows you to resize your servers in just a couple of clicks.

00:18:10   I've done this myself many times.

00:18:11   It's been awesome.

00:18:12   In fact, I'm actually doing one later today.

00:18:13   I'm running out of disk space on my backup server.

00:18:15   I'm going to just go and resize it later today.

00:18:16   It's going to take about 10 minutes, most of which I'm just waiting for it to complete,

00:18:19   and it's going to be great.

00:18:20   I've done it so many times.

00:18:21   I know exactly how it's going to go, and man, I love Linode.

00:18:24   And all of this is also manageable if you want via the command line through their API

00:18:27   as well, if you don't want to use their interface or if you want to automate large tasks.

00:18:31   All of Linode's pricing tiers feature hourly billing with monthly caps on all plans and

00:18:35   add-on services like backups and node balancers.

00:18:39   So Linode, they also have some awesome new pricing options.

00:18:42   You can now get a server with one gig of RAM for just five bucks a month.

00:18:48   I can't believe how cheap servers are.

00:18:51   This makes me so happy because servers used to be so expensive.

00:18:54   And when I was starting out, the idea of five bucks a month for a server with a gig of RAM,

00:18:59   oh my god, I paid so much more for so much less back then, and you can go all the way

00:19:02   up to 16 gigs of RAM for only 60 bucks a month.

00:19:06   Oh man, their prices are good.

00:19:09   And their two gig plan now includes 30 gigs of storage, and that's just 10 bucks a month

00:19:14   for two gigs of RAM, 30 gigs of SSD storage.

00:19:17   Linode continues to offer more and more awesome options to meet your server needs.

00:19:21   Across the board, they're offering twice the amount of RAM that you get elsewhere.

00:19:25   As listener of this show, if you sign up at linode.com/radar, you'll not only be supporting

00:19:30   us but you'll also get $20 towards any Linode plan, and with a seven-day money-back guarantee,

00:19:35   there's nothing to lose.

00:19:36   So go to linode.com/radar to learn more, sign up and take advantage of that $20 credit or

00:19:41   use the promo code radar2017, that's radar2017, just like the year, I hope, at checkout.

00:19:48   Thank you so much to Linode for supporting this show.

00:19:50   All right, and the next area that I've been recently battling through with the same app

00:19:56   that I think is interesting to talk about is the concept of iOS deprecation.

00:20:03   So the app when I launched it, when I started this project, came up, I think there was something

00:20:08   on the order of 80 deprecation warnings, which is certainly an indication of my, I guess,

00:20:17   ability to keep on top of things.

00:20:21   My favorite was I had one deprecation warning that said, I think it had something to do

00:20:25   with the audio session APIs, and it was the set delegate method, and it said, "This method

00:20:30   was first deprecated in iOS 6."

00:20:34   So I'm a little behind, and it's amazing that Apple has been very kind to me, and they deprecated

00:20:41   it in iOS 6, and it still continues to work just fine in iOS 10, but deprecation is probably

00:20:48   worth just saying briefly what I'm talking about.

00:20:50   So this is the, Apple will take a particular API, a particular method, and say, "We are

00:20:58   deprecating this method," which means we have something new or we have something different

00:21:02   that we want you to use, and usually the deprecation warning will even be nice enough to say, "You

00:21:06   shouldn't be using this method anymore.

00:21:08   You should be using this one," and point to it.

00:21:12   And what they're saying is, at some point, indistinctly in the future, this method is

00:21:17   going to go away, or it'll stop working correctly, or whatever, and this is their polite way

00:21:22   of saying, "You need to change your code to work with the new stuff."

00:21:31   Deprecation I feel like is one of these funny things where it reminds me a lot of the Hemingway

00:21:35   quote about going bankrupt, where you go bankrupt two ways.

00:21:39   You go bankrupt gradually, and then suddenly, because that's essentially what the deprecation

00:21:44   warnings are.

00:21:45   You very gradually have this building up, like I have these deprecation warnings that

00:21:48   have been building up over time, and they're like, "I should be changing things.

00:21:51   I should be improving things," and then one day suddenly, it'll just stop working.

00:21:55   iOS 11 will come out this June, and then suddenly I no longer have the choice and the flexibility

00:22:00   and the ability to just like, "Oh, I'll get to that whenever I want to."

00:22:05   It's like, "No, suddenly I have to do it.

00:22:07   The app will no longer compile."

00:22:11   So it's a really tricky thing, though, because the thing that I always struggle with with

00:22:15   deprecation is it's the issue of when to do it, because the code works as it is right

00:22:23   now.

00:22:24   And that is one of the hardest things, I think, for me to be able to justify and work, is

00:22:29   I'm working on an app.

00:22:30   Here's this function that works correctly, like functionally has been tested, has been

00:22:35   out in customer hands for, in this case, for years and years, and it works, it's reliable,

00:22:41   all the bugs have been worked out.

00:22:42   For me to come along and to change it when it is working now always feels a little bit

00:22:48   precarious.

00:22:50   And while I know that it's a bad thing to not, in theory, or I should probably be doing,

00:22:55   is as soon as I see a deprecation warning, I need to be going in there, changing it right

00:22:59   at that very moment.

00:23:01   But at the same time, APIs are most buggy when they're new, and there's lots of other considerations

00:23:07   that come into that that mean that sometimes I want to wait, but at the same time I don't

00:23:12   want to end up completely bankrupt at the end, where suddenly I'd have no choice and

00:23:15   I have to change everything at once, and then I'm changing, it's even more dangerous than

00:23:19   changing little things along the way as probably changing 80 warnings as I am right now.

00:23:25   That's a bit more problematic.

00:23:26   But that balance and that tension, I find, is really awkward and complicated to deal

00:23:30   with.

00:23:31   - I mean, the way I deal with deprecation is kind of like what I was saying last week

00:23:35   about Swift adoption, which is like, it's the kind of thing where a deprecation is basically

00:23:42   a note, it's adding something to your to-do list, that you're going to need to deal with

00:23:46   this.

00:23:47   And like most to-do items, I think practitioners of GCD and similar systems, there's almost

00:23:53   always a rule in those systems that says something along the lines of, if you can do something

00:23:58   really quickly, just do it right now and just get it done.

00:24:01   And if not, then you schedule it and prioritize whatever else.

00:24:04   Sorry, I'm not an expert in those systems, but I know there's usually the rule that says

00:24:08   if it's easy, just do it now.

00:24:10   And that's kind of how I look at deprecations and things like new languages, new APIs, new

00:24:17   frameworks, that the easy stuff, deprecations are usually really easy to get around.

00:24:22   Usually it's like, oh, there's now a new method that has an options parameter or a callback

00:24:26   or something else, it's like, usually the way to deal with it is very quick, you can

00:24:31   do it in a couple minutes at most and you're done and then that's it.

00:24:35   And the problem, as you said, it's like, similar to what I was saying last week about Swift,

00:24:38   about like, you know, at some point in the future there's going to be some API or some

00:24:42   policy or something that requires Swift and then like I'll be forced to learn it right

00:24:47   then, which is kind of not on my own time, not on my own schedule.

00:24:51   And it's similar with deprecations, like at some point you're going to be forced to do

00:24:55   this and so you're better off doing it when you have time on your own terms, on your own

00:25:01   schedule as opposed to in the future being forced suddenly, you know, oh, now I need

00:25:06   to do this instead of what I was supposed to be doing today.

00:25:09   It can also happen with app review, like there could be a time where you need to submit a

00:25:13   critical bug fix to your app and that happened to be the day that they started enforcing

00:25:22   this particular deprecation warning that you're hitting in app review and all of a sudden

00:25:25   now you can't even submit this bug fix that you need to submit because you can't submit

00:25:29   anything until you fix this deprecation.

00:25:31   And so it's a kind of thing where like most deprecations are easy to fix.

00:25:35   If you can fix it, just do it now because there will never be a good time later to do

00:25:39   it.

00:25:40   >> Yeah, and I think the only thing that's also complicating into this is the broader

00:25:45   the range of iOS versions that you support, the more complicated this becomes, I think.

00:25:51   >> That's true, yeah, that's true.

00:25:52   >> Because one of the things, like part of the reason why this app has such a wide range

00:25:57   of deprecation warnings is, you know, a lot of them were when I changed the minimum supported

00:26:02   version from iOS 8 to iOS 9, you know, then suddenly all the things that were deprecated

00:26:07   in iOS 8 become deprecated because before that they weren't deprecated.

00:26:11   In fact, they were essential because if I installed the app on a device that runs iOS

00:26:17   8 and used the new API, it won't know anything about it and it won't work.

00:26:22   And so like this is something too that it's probably worth keeping in mind.

00:26:27   Like I think something that I've learned from this experience is that even if I'm not dropping

00:26:31   support for the older version right away, like when iOS 11 comes out this June presumably,

00:26:39   I'm unlikely to drop support for it right away for this particular app.

00:26:42   For some of them I might, but for this one I probably won't.

00:26:45   What I probably should do though is set the minimum version to iOS 11 and see what is

00:26:52   deprecated as a result, to get a sense of how big this hole I'm digging myself into

00:26:59   by keeping these things around.

00:27:01   And I mean, I will say to Apple's credit, like they are very good about this.

00:27:04   Like it is kind of crazy to me that how much work they seem to put into backwards support

00:27:11   and backwards compatibility, the fact that I can run a really old version using modern

00:27:17   Xcode is quite a testament to them.

00:27:19   And similarly, the fact that I can build, I don't know how far back I can go with the

00:27:24   modern version of Xcode, like if I could support iOS 4 or iOS 3.

00:27:28   I'm sure at some point Apple cuts it off, but the fact that they go back a couple versions

00:27:35   at least and cover the vast majority of users as a result is certainly a testament to the

00:27:41   engineering priorities and making sure that their devices, that you don't have these very

00:27:46   abrupt and very unexpected cutoffs, which I'm sure it takes effort and time on their part,

00:27:53   so they don't have to do that, but I'm glad that they do.

00:27:57   And I think the last thing that I've also just been thinking about as I've been going

00:28:01   through this is something that I notice when I'm going through these kind of cleaning passes,

00:28:06   and going through and cleaning up something, is it's always funny when you're going through

00:28:11   an old code base and you find methods that, like I've been noticing this when I'm, mostly

00:28:16   when I've been fixing deprecation warnings, like I find this method that has a deprecation

00:28:19   warning in it, I go into that method and I'm like, "Huh, I wonder where this is used."

00:28:22   And I do like, you know, a search all for the method signature, and it's not used anywhere.

00:28:28   Like it's kind of a funny thing where it's like there's this method that, you know, there's

00:28:33   all this code in the app that has no references to it anywhere because it existed at some

00:28:39   point, and then I rewrote the thing that was referencing it, and then it just sort of accumulates

00:28:44   over time.

00:28:45   And I went into AppCode, which is the IDE that JetBrains put out, which is like, I don't

00:28:51   use it for, I wouldn't ever use it for development, like I really like the way Xcode works, but

00:28:55   it has a bunch of these kind of refactoring tools, and like I found that it said, I think

00:28:59   there was 426 methods that were unused in the application, but it's just like, it's

00:29:04   the nature of, it's existed since, you know, spring of 2009.

00:29:09   And so I'm going through and cleaning this up, and I think it's, again, it's just one

00:29:12   of these things that's reminded me of like, it's probably good as I'm going to be a bit

00:29:16   more thoughtful about this, that when I'm removing, you know, as I'm making changes,

00:29:20   if I'm removing the use of a function for the last time, I should probably, you know,

00:29:25   clear it out there and remove the amount of code in my project because it's just increasing

00:29:30   the odds that, you know, there's going to be these lingering and weird bugs.

00:29:34   Like the less code in my app, the less bugs there are, almost certainly, so it's just

00:29:37   like a good reminder to me that I should be cleaning things up a little bit more.

00:29:40   All right, we're out of time this week.

00:29:42   Thank you everybody for listening, and we'll talk to you next week.

00:29:45   Bye.

00:29:45   [