Sagacity posted:

Well, I assume it's not literally the text "client" and "secret"?

But isn't it the same when you use a key for accessing the Google Maps API from your website?

Tei posted:

Writing bad code is like spitting straight up. It may feel good for a second, but then the spit lands on your face again.

Writing bad code would be like get asked to wear adult diapers with poo poo in it, and then travel like that from New York to San Francisco.

Is real bad for your mental health.

QuarkJets posted:

Can't you find a thread in gbs or pyf to post your kinks in?

Hammerite posted:

Everyone has a certain number of lines of bad code in them that they have to write before they can write good code

For some people it is more and for some people it is less but everyone has to get the bad code out of their system before they can start producing good code. I sincerely believe it

Xarn posted:

Believing that you can "get the bad code out of your system" is how you get into this thread.

Xarn posted:

Believing that you can "get the bad code out of your system" is how you get into FOSS maintenance.

Xarn posted:

Believing that you can "get the bad code out of your system" is how you get into this thread.

quote:

Sorry, but we’re having trouble with signing you in.

AADSTS50020: User account 'myname@mycompany.com' from identity provider 'https://sts.windows.net/some-guid/' does not exist in tenant 'Microsoft' and cannot access the application 'some-other-guid'(RedirectionUxProd) in that tenant. The account needs to be added as an external user in the tenant first. Sign out and sign in again with a different Azure Active Directory user account.

Hammerite posted:

• identified a minor issue with a part of Azure DevOps's web interface and decided to request an improvement
  
• googled "azure devops bug"
  
• found this page: https://azure.microsoft.com/en-gb/support/devops/
  
• clicked the link that says "report an on-premises bug"
  
• the link takes me to a Bing search page
  
• tried a shortened form of the URL. that doesn't send me to the Bing search page, but instead results in this:
  

loving hell. never mind

Tei posted:

I have never like how Microsoft does things. It appears the systems are a bunch of disjointed products every one build by different people, nobody really care about the user. The global experience is bad because the systems separate enough you can fall in the cracks, to be never seen again.

Volmarias posted:

I have bad news for you about every software company

Hammerite posted:

identified a minor issue with a part of Azure DevOps's web interface and decided to request an improvement

DaTroof posted:

Not believing it is how you become unemployable at 40.

Mega Comrade posted:

Xarn posted:

??

Presto posted:

You never get the bad code out of your system. You just get better at making your bad code standards-compliant.

Tei posted:

I have never like how Microsoft does things. It appears the systems are a bunch of disjointed products every one build by different people, nobody really care about the user. The global experience is bad because the systems separate enough you can fall in the cracks, to be never seen again.

Tei posted:

I have never like how Microsoft does things. It appears the systems are a bunch of disjointed products every one build by different people, nobody really care about the user. The global experience is bad because the systems separate enough you can fall in the cracks, to be never seen again.

Mega Comrade posted:

One of my friends who is a designer just messaged me, thought I'd share.

canis minor posted:

This reminds me when at my old place devs put in a system that would strip out javascript comments using regexp - everything beginning with // to end of line and everything between /* and */. Unfortunately this meant that hardcoded urls like http://foo.bar would be stripped out and http: would be left to be loaded. The solution? Replace hardcoded url with http:\/\/foo.bar.

Tei posted:

I have never like how Microsoft does things. It appears the systems are a bunch of disjointed products every one build by different people, nobody really care about the user. The global experience is bad because the systems separate enough you can fall in the cracks, to be never seen again.

Blue Footed Booby posted:

The thing that gets me about Microsoft is how often it seems like the different products aren't just made by totally disconnected groups, but by groups that are actively hostile to one another. The wild swings in the direction of Windows feels like the result of some crazy pre-unification Japan style civil strife.

ultrafilter posted:

I think that officially that's less true than it used to be, but a lot of long-time employees still have that mentality.

Ola posted:

Perhaps they're like European nations, different cultures being friendly across borders hammered out with blood.

SirViver posted:

That reminds me of an old version of the Oracle Managed Data Access (.NET) client that stripped out comments (for unknown reasons) by removing everything after a "--" in a given line with absolutely zero SQL syntax parsing and also stripping out all /* */ comment blocks.

The first had the hilarious consequence of obviously breaking queries such as this:

SQL code:

  SELECT foo 
    FROM bar 
   WHERE xyz = '-- Some Text --'
ORDER BY foo

Which was then executed as:

SQL code:

  SELECT foo 
    FROM bar 
   WHERE xyz = '
ORDER BY foo

The second one made it so you were actually unable to include query optimizer hints, which are - you might've guessed it - placed inside queries via comment blocks a la:

SQL code:

  SELECT /*+ HINT_GOES_HERE */ foo 
    FROM bar

Presumably those would've also broken queries if /* was placed inside a string literal, but I can't remember if I ever tested that.

I think they later "fixed" this by only removing comments when they were placed at the start of the line (so it might still partially happen if you format queries including optimizer hints in a certain way). Why they do this to begin with I have no idea. Maybe some people include kilobytes worth of comments in ad-hoc queries that they attempted to strip out to save network bandwidth?? In either case, just futzing around in a query like that with zero regards for syntax is truly amateur hour level poo poo. Maybe they made the interns code their .NET client, who knows.


Another thing I remember is that some regex they execute on the query had (maybe still has?) terrible performance if you happened to include a bunch of consecutive whitespace anywhere in it. I found that out because some query processing code of a third party framework that we use for supporting our old legacy application (that was machine-ported from a now long dead application language to .NET) happened to strip certain custom pseudo-SQL syntax by overriding it with spaces instead of removing it from the string - probably to not chance loving up any converted code that relied on precomputed string indices, so technically a wise choice. The result of that was that a query including e.g. 500-1000? spaces anywhere inside it took like 4 seconds of 100% CPU time on one core doing nothing but processing that regex. Granted, that's something I fault Oracle less for as that's a really weird scenario to begin with. But then again why do they insist on doing so much "client-side" processing at all? Just let the friggin Oracle server deal with that poo poo, so you don't unnecessarily eat CPU on every query you execute.


Oh yeah and the hilarious default behavior to not bind DB parameters by name but by index instead. You heard right, by default the order you add the parameters to the command must match the order they appear in the query in. That sure was a headscratcher why random queries suddenly broke after converting from the native to the managed client. Thankfully that was easily fixed with a single setting. The reason they do that btw is that they don't know what hash lookups are and if your query contains a larger amount of DB parameters (e.g. the generated insert statement for a wide record) then their internally executed linear array lookups to find the parameter by name will start taking up a significant amount of time when you do repeated inserts. So to avoid that let's just default to index instead, problem solved. If you use names it's your own drat fault. Oh but of course instantiating a dictionary for fast lookup when needed would be such a huge overhead... better save that so we have more time executing multiple regexes per query to do who-knows-what.


TL;DR Oracle is poo poo.

Nth Doctor posted:

I literally wrote my thesis on solving the problem of properly stripping comments out of SQL Queries.

ultrafilter posted:

https://twitter.com/mycoliza/status/1446238022949695521