kriegor: (xan)
2019-01-02 10:48 am

dd 3

Here's to 2019.

Been a couple minutes. My understanding of the Windows API is starting to come together, and it's simultaneously hyper-enriching and terrifying-- I didn't know it was that bad, but I didn't know it was that simple, either. The sensation is skin to getting just enough study in where, looking at a foreign language, you're understanding 50% of the words where in the past you got 5-10% at best.

To something-like-summarize...

The windows API -- as in, the code you write, functions you call, kernel bits you prod, etc. to spawn windows and get them talking -- has not changed in many backwards-incompatible ways in something like thirty years. Windows 10 seems to have tried (and to some extent failed) to deprecate Win16, ie. the API calls specific to Windows 3.1 and prior, but it's not comparable to Apple's deprecation of PowerPC.

New overlays (eg. MFC), new 'translations' (eg. every library portable to Windows), new frameworks (eg. .NET), but nothing excised.

Considering part of the reason I cling to C so tightly is because of Linux-and-co's long-lived tradition with it, this makes me much more comfortable writing code for Windows. If still a little horrified.

Mostly because Windows' development tradition is in C++, not C. The peppering of 'object' and 'component' everywhere is a hint. And C is largely an afterthought these days. (Often a contemptuous afterthought lending to my overall silence.) WinRT, the 'new' API going forward, doesn't officially support C at all, though you can still make it work.

Ah well. With the general allergy to the console I imagine it's a bit like coding for the Mac was back in the 90s. At any rate, nothing really changes for anyone else other than seeing a spare 'cygwin.dll' lying around; one of my priorities is giving people a single EXE or the like to click/drag/interact with, because that makes the most sense on Windows, and while Node's 'uhh run the engine and...' approach on Windows violates this, dynamic linking does not. (OSX is much better-suited for Node's... situation, when it comes to packaging and distributing Electron apps and such. Yes, I know you can 'compile' Node more tightly. It also takes an extra toolchain whereas compiling C is a built in functionality. Also, it's Node.)

I'm making a few smaller applications before I get back to the main meat. Notably, a little tool with sockets so that I can be pinged without being aggravated (TCP/IP code, JSON in C, tray applications...), and an egg timer that doesn't suck (timers, shaped windows, multithreading...). I have been pressed into using hourglass for the latter since my OS hop and ... by some quirk of my system it takes nearly 30 seconds to open every time. I don't dare speculate which quirk that is, knowing what I know now about Win32.
kriegor: (Default)
2018-12-20 12:13 pm
Entry tags:

uc2e dd 2

Somehow I've only known about compression algorithms in the vaguest possible sense until this week.

zlib and friends )

Theoretically I'd like to get said library working today but I have to go for groceries and have no data on how foggy I may be when I get home; really wish I had the affordance to take a caffeine tolerance break. All that needs to be done is to 'apply' it to a byte buffer I'm already reading, however my still-extant questions make me sort of freeze up when I look at the problem (it's a couple of nested loops deep) as without this information I'm prone to further unnecessary complication. I don't really do well with unanswered questions or unclear behavior.
kriegor: (xan)
2018-12-19 11:25 am
Entry tags:

uc2e dd 1

Disclaimer/explanation/etc: This is a code diary and most of this won't make sense; I'm largely talking to myself. If you like to watch other people flay their own brain in real time though (I sure do!) knock yourself out.

I have this rampant, rampant overdesigning problem-- but, I think, it takes a slightly different form than most overdesigning problems, at least the ones I observe-in-public. It's not that I get lost among the trees. I hate deep detail diving, actually. It's that I can see the entire shape of a problem but consistently lack the muster, rapport or simple trust to delegate anyone other than myself-- and 'myself' is consistently smaller than projected (a history of mania distorts this, it does).

prelude )

post-prelude )
kriegor: (xan)
2018-12-11 01:38 pm

Attempting to replicate Tumblr is a mistake.

I have been meaning to write something about this for years; my perfectionism and vague goals led to a bunch of scattered notes. Now it's either too late, or more relevant than ever. So: I'll forgo the notes entirely and make this argument from memory-- I've certainly retread it enough...

Let's not be coy-- transplanting from any other social media site to DW (or, really, between any two nonfederated sites at all) is a culture hop. And when you ask for something that is familiar to do, and are told 'that's not the way we do it here', a bit of hostility is natural. I think 'tough, we don't have that' is a pernicious response, if only by sake of omission.

An example-- one I've already seen in the wild a few times now-- is looking for a 'like' function. Personally, I don't like 'likes'-- I think they encourage obnoxious or degrading behaviors, like beancounting-- and I definitely don't think they fit with Dreamwidth's format, which funnels users toward a more direct response in the form of the comment box. But even though I may not like them, I am not so naive to imagine they serve no constructive purpose under any conditions. I've seen them used to express 'I hear you' in contexts where that was helpful and a myriad other -helpful- things that would not really be elucidated by a worded reply.

I am neither so naive to imagine anyone else ought to share my opinion simply because it's my opinion. That's culture for you, though.

There is something valuable in the confluence of social features Tumblr has, including that one-- it's easy to post, disseminate, acknowledge and collect images, which encourages formats of creativity and sharing you can't really find anywhere else. Its tagging system is powerful and double-edged and handles 'discovery' better than nearly any other platform I've seen-- if you want something more niche, you use a more niche tag, and you can find those -through- less niche ones by association. And it's -quick- by merit of how visual most of the website is-- and how its format encourages that.

If you're paying attention, you may notice that most of these points could be easily rephrased as criticisms.

I would argue, in spite of all this, Tumblr's ownership history, incompetent programming, flippant approach to user privacy, and design practices that are at best shortsighted and at worst predatory, offsets much of the potential value its strengths would offer. It could be so much more. Perhaps if you aren't affected by one or more of the downsides-- or aren't aware of the loss, like so many people I've met who had no idea that real privacy and authorization locks in online journals were the norm pre-tumblr, not just an unspoken accord that other peoples' personal reflections might not be your business-- the potential strengths look that much better.

Therein lies the danger of trying to succeed Tumblr by replicating Tumblr; if the party doing so doesn't have a firm, grounded and wide angle view of what Tumblr did wrong, as well as what it did differently from its predecessors (if other social media sites can really be called 'predecessors' to it-- I mostly refer to 'sites it siphoned many of its users from), then the only model they'll have to go off of is a lemon.
kriegor: (xan)
2018-12-11 09:28 am

Some notes on sussing out Tumblr alternatives.

If I were you, I would be skeptical of:


  • Websites that don't have weird ALL CAPS BLOCKS in their TOU. While not 100% accurate, and while their presence may be vaguely intimidating, this tends to be a hint that a law professional was not involved, which, at the moment, is a bad thing.

  • Websites stating 'being a safe space' in their mission. What is safe for someone else, by design and not mistake, will not be safe for you. In practice, this tends to be a sweetly-worded 'we do what we want' dispensation for administration.

    • As an extension of this, websites with TOUs that are casually worded, doubly so if they aren't backed by a 'legalese version'. Yeah, sure, it's less intimidating-- until you actually have to use it...

  • Websites that have taken venture funding. This is usually a hint that 1) they are beholden to someone, and 2) they aren't turning the kind of profit that would keep them afloat were this 'someone' to get antsy. To research this, queries like '[website name] funding' can be useful.

    • If you'll be using a website for anything even slightly consequential, it's always wise to know where their money comes from, VF or not.

  • In light of the volume of users leaving Tumblr, for the next few months I'd avoid websites with a single administrator, as handling a huge, sudden influx of users is both financially and personally stressful and those setups seem most likely to fold under the pressure. (As vital as I believe small, individually-run spaces may be going onward, I challenge you to find a lone, untrained person who can safely handle a firehose.)

  • Anything that ends in "-blr", the newer the worse. I've seen quite a few attempts to 'cash in' on the shutdown, some well-meaning, some not-so-much.

  • Websites that are a subsidiary of any larger corporation; given what happened with Tumblr leading up to the current TOS change, this probably goes without saying.



I would argue the last bit is why we haven't seen much roiling from Twitter yet. I guarantee you it's coming, though.

I put my emphasis on skeptical, not avoidant. It's not to say I wouldn't use them; just that I would use them slathered in the awareness of my own risk. Say, Mastodon -- not every instance is going to have a legally-vetted TOU, nor, I argue, should they -- albeit it does seem awfully like we're getting close to it being functionally mandatory, even for lone administrators who could never afford it, huh?

Huh...
kriegor: (krieg)
2018-12-10 07:35 pm

Try folding THAT into your passport, why don't you.

It's time to dust off this good old bad book. Tumblr shutting down funnels you all-so-many back or over here; all the better for me. And let's be acid frank-- it /is/ shut down, for all intents and purposes. That bot needs therapy.

Welcome folks, new and-or old. Worry not; the Livejournal codebase and I have had an understanding (including this site in particular) since the year had headlights in it. Let me know if you get lost. I'll be the first to tell you it's obtuse, but I've found a skepticism of convenience is healthy, to an extent; maybe you'll find that true, too.

I'm going to try using this as a coding diary for a little bit. I've got awful conniptions when it comes to computers-- the discipline's horse is tied to the ego's post and they're both putting down roots, which sucks. I could use the implicit audience.

+ Oh, yeah. I get to set up a new smut filter, too! How quaint!