** UnhandledExceptionEventHandler :: OFFICIAL LC / CHATTER THREAD ** - Page 1405 - Computer Technical Help

Imo if you don't push every commit eventually you will lose work. It might take a long time but it will happen. Since we squash and merge to master I don't care about feature branch history anyway.

with git reflog you can get every commit on your local. no matter if you hard reset (ie undo) it.

Quote

09-08-2018 , 11:09 AM

#35102

jjshabado

Carpal Tunnel

Join Date: Jul 2006 Posts: 22,732

Grue’s point though is that you could lose your local disk and any work you just have locally.

I guess he’s probably right but it doesn’t seem like a fight worth having given how incredibly rare this would be.

Quote

09-08-2018 , 01:26 PM

#35103

KatoKrazy

Pooh-Bah

Join Date: Jan 2009 Posts: 4,533

I used to work for an HDD company. I backup to remote daily if that tells you anything

Quote

09-08-2018 , 04:10 PM

#35104

suzzer99

Save the Cheerleader, Save the World

Join Date: Nov 2005 Posts: 99,984

Can someone explain what this means? Like you instantiate an object once, then you can't ever change the value of any property - instead you have to create a new object?

Ok I guess so. https://en.wikipedia.org/wiki/Immutable_object

Quote

09-08-2018 , 04:28 PM

#35105

Larry Legend

Celtic Pride

Join Date: Jul 2009 Posts: 42,920

Yes, look into immutabe js.

This is kinda like typescript in that it seems kinda annoying at first but its worth doing and avoids tons of bull****.

It is a pattern I originally learned from Redux and then went down several rabbit holes and now if I ever see "object.foo = x" I immediately cringe, or the even more cringeworthy "delete object.foo".

Quote

09-08-2018 , 04:40 PM

#35106

suzzer99

Save the Cheerleader, Save the World

Join Date: Nov 2005 Posts: 99,984

I feel like a case where you get an object back from the DB and want to delete a couple of sensitive properties before sending on to the browser should be an exception.

Quote

09-08-2018 , 05:40 PM

#35107

heehaww

Pooh-Bah

Join Date: Aug 2011 Posts: 5,081

Any opinions on this? - Tuple Spaces (or, Good Ideas Don't Always Win)

Quote:

It's as if the metric system had failed, and we had to do physics with foot-acres and what-not.

Fair assessment?

Quote

09-08-2018 , 06:18 PM

#35108

goofyballer

Carpal \'Tunnel

Join Date: Jun 2005 Posts: 70,719

Quote:

Originally Posted by suzzer99

gaming_mouse (don't think he's been around much lately?) used to espose things like this too, I think.

I'd be curious to see a large codebase full of immutable objects in action, anyone know of any on Github?

Quote

09-08-2018 , 06:20 PM

#35109

suzzer99

Save the Cheerleader, Save the World

Join Date: Nov 2005 Posts: 99,984

Well react is basically built on that concept right?

Quote

09-08-2018 , 08:46 PM

#35110

Craggoo

banned

Join Date: Sep 2006 Posts: 12,644

Quote:

Originally Posted by suzzer99

Well react is basically built on that concept right?

I can say that Elixir is based on that idea. You never modify an existing object. You create a new one.

Quote

09-09-2018 , 12:53 AM

#35111

Victor

Carpal \'Tunnel

Join Date: Jul 2003 Posts: 71,803

Quote:

Originally Posted by suzzer99

I feel like a case where you get an object back from the DB and want to delete a couple of sensitive properties before sending on to the browser should be an exception.

create a new object?

Quote

09-09-2018 , 01:05 AM

#35112

suzzer99

Save the Cheerleader, Save the World

Join Date: Nov 2005 Posts: 99,984

Why? You're not gonna use the object for anything else, it dies as soon as the response is passed anyway.

Quote

09-09-2018 , 11:05 AM

#35113

Larry Legend

Celtic Pride

Join Date: Jul 2009 Posts: 42,920

In a robust system with sufficient logging it lives on somewhere.

Quote

09-09-2018 , 12:44 PM

#35114

Victor

Carpal \'Tunnel

Join Date: Jul 2003 Posts: 71,803

Quote:

Originally Posted by suzzer99

Why? You're not gonna use the object for anything else, it dies as soon as the response is passed anyway.

bc those are the rules. in other words, bc thats how it passes code review.

Quote

09-09-2018 , 03:00 PM

#35115

suzzer99

Save the Cheerleader, Save the World

Join Date: Nov 2005 Posts: 99,984

Quote:

Originally Posted by Larry Legend

In a robust system with sufficient logging it lives on somewhere.

You also don't want sensitive data going into logs.

Quote

09-09-2018 , 03:17 PM

#35116

blackize5

old hand

Join Date: Sep 2012 Posts: 1,358

Well you could just ask only for the data you need. Or take your model object and use a serializer or decorator to map to a new object without the sensitive fields.

You can see how starting to add special cases where mutation is ok begins adding additional cognitive load. Much easier for the language to just not support it or to disallow with a linter or whatever.

Quote

09-09-2018 , 03:21 PM

#35117

suzzer99

Save the Cheerleader, Save the World

Join Date: Nov 2005 Posts: 99,984

These objects come from back end services we don't control. How is a model object or decorator less cognitive load than deleting a couple properties before sending JSON to the browser?

Quote

09-09-2018 , 04:47 PM

#35118

blackize5

old hand

Join Date: Sep 2012 Posts: 1,358

Well it doesn't make a huge difference for that exact use case. But in the general it's very easy to go from just deleting a sensitive key to just deleting a sensitive key, just translating a value, just adding a default value...

So in my mind holding to the rule even in the simplest case is helping to prevent death by a thousand cuts.

I don't really have a strong opinion about this in languages that support and encourage mutation, but I do appreciate elixir for eschewing it.

Quote

09-09-2018 , 09:39 PM

#35119

Larry Legend

Celtic Pride

Join Date: Jul 2009 Posts: 42,920

Quote:

Originally Posted by suzzer99

You also don't want sensitive data going into logs.

Yea true logging wasnt the best example for that.

But in practice I've always just copied the object by picking the keys that you want in the new object.

Quote

09-09-2018 , 10:19 PM

#35120

candybar

Pooh-Bah

Join Date: Aug 2011 Posts: 4,218

Quote:

Originally Posted by suzzer99

These objects come from back end services we don't control. How is a model object or decorator less cognitive load than deleting a couple properties before sending JSON to the browser?

I don't necessarily care about enforcing immutability per se but I think in this case constructing a new object with just the fields you need is the right thing to do anyway since you want to control exactly what fields are sent to the client. Otherwise backend services can add fields to the payload and you will end up automatically including them.

Quote

09-10-2018 , 12:22 AM

#35121

suzzer99

Save the Cheerleader, Save the World

Join Date: Nov 2005 Posts: 99,984

Well the counter is that maybe you want to include them. So now instead of changing code in two places (api, client) to add new fields you have to change it in 3 places (api, client, middleware). I think for some cases a field blacklist instead of whitelist is fine - but it's obviously highly situation dependent.

That's the only time I can ever think of using delete.

Quote

09-10-2018 , 07:12 AM

#35122

maxtower

Carpal \'Tunnel

Join Date: Sep 2005 Posts: 7,996

I think you're making this too complicated. Instead of thinking every time you want to change an object, you just make a copy of that object with the properties changed. Some languages make this easier than others.
For example, in kotlin (java makes this more verbose) you can just say:

Code:

data class DataState(
  val user: User,
  val goldStars: Int,
  val isOnline: Boolean
) { 
fun setUser(user: User) = copy(user = user)
}

Then to make a new object of the DataState:

Code:

val newDataState = oldDataState.setUser(newUser)

in the above example, oldDataState is immutable and doesn't change. Anything referencing it always points to the same values. Immutability matters a lot more in multi-threaded applications (but thats most of them) because side effects won't occur as a result of the data changing

Quote

09-10-2018 , 12:17 PM

#35123

suzzer99

Save the Cheerleader, Save the World

Join Date: Nov 2005 Posts: 99,984

Yeah you can do that for sure. But delete is still less code, and I just don't see an advantage to making an object immutable for immutability's sake in a node/express app - where literally the whole framework is based on muting objects (adding to res.locals to pass along through the express middleware chain).

Quote

09-10-2018 , 07:53 PM

#35124

suzzer99

Save the Cheerleader, Save the World

Join Date: Nov 2005 Posts: 99,984

Btw I made a greasemonkey (now called tampermonkey) script to scroll pages back to the correct first unread post - after the tweets load inline and screw up the scroll position.

Code:

// ==UserScript==
// @name         New Userscript
// @namespace    http://tampermonkey.net/
// @version      0.1
// @description  try to take over the world!
// @author       You
// @match        https://forumserver.twoplustwo.com/*index*.html
// @grant        none
// ==/UserScript==

(function() {

    setTimeout(() => {
        $('html, body').animate({
            scrollTop: $(window.location.hash).offset().top + 'px'
        }, 100, 'swing');
    }, 3000);

})();

On Chrome tampermonkey is an extension, not sure about the other browsers. I set the timeout to 3 seconds which is a total guess. Seems to work ok so far.

It also should be easy to wire up to a hotkey to scroll back to the correct position instead of 3 second timeout if that's preferred.

Ideal would be to capture the event when all the tweets are done loading. But aint nobody got time for that.

Quote

09-10-2018 , 08:17 PM

#35125

Grue

Pooh-Bah

Join Date: Mar 2004 Posts: 5,667

I don't think they're globally exposed anyways? Theres no xhrs so it has to be iframes. No idea how you'd get in there from the console to get exposed to things happening in there.

Maybe a better solution is to count the iframes and move the window up iframe count x 400px or something?

Quote

Page 1405 of 1603

First

405 905 1205 1305 1355 1385 1395 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1415 1425 1455 1505

Last

Post Reply Subscribe

...

Page 1405 of 1603

First

405 905 1205 1305 1355 1385 1395 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1415 1425 1455 1505

Last