Beating another dead horse.
  •  lemmyvore   ( @lemmyvore@feddit.nl ) English
    24•3 months ago

    These days I follow a hard heuristic: Always use synthetic keys for database tables.

    And the way to follow this rule is fairly simple, but it has a few twists.

    For internal use, the best and most common key (in a relational database) is an auto-generated incremental sequence. But it it ok to use it externally? – across databases, across types of data storage, across APIs / services etc.

    It’s tempting to refer to the sequence number in API calls, after all they are going to that particular database and are only going to be used with it, right? Well not necessarily; the database and the code powering the API are different systems, who says there won’t be other apps accessing the database for example.

    The current OpSec school of thought is that sequence keys are an internal database mechanism and sequence numbers should only be used for internal consistency, never used as external references (even for the “local” API).

    Sequence keys also don’t offer any way to deal with creating duplicate data entries. If you’ve been around for a while you’ve seen this, the client sends the same “create” request twice for whatever reason (UI lets user multiple-click a button, client assumes timeout when in fact it had gone through etc.) Some programmers attempt to run heuristics on the data and ignore successive create attempts that look “too similar” but it can backfire in many ways.

    An UUID is pretty much universally supported nowadays, its designed to be unique across a vast amount of systems, doesn’t give anything away about your internal mechanisms, and if you ask the client to generate the UUID for create requests you can neatly solve the duplicate issue.

    Do keep in mind that this doesn’t solve the problem of bijection across many years and many systems and many databases. An entity may still acquire multiple UUID’s, even if they’re each individually perfectly fine.

    There can also be circumstances where you have to offer people a natural-looking key for general consumption. You can’t put UUID’s on car plates for example.

    •  CasualTee   ( @CasualTee@beehaw.org ) 
      4•3 months ago

      On the topic of exposing sequence number in APIs, this has been a security issue in the past. Here is one I remember: https://www.reuters.com/article/us-cyber-travel-idUSKBN14G1I6/

      From the article:

      Two of the three big booking systems - Amadeus and Travelport - assign booking codes sequentially, making brute-force computer guesswork easier. Of the three, Amadeus, through its web portal CheckMyTrip, is especially vulnerable, Nohl said.

      The PNRs (flight booking code) have many more security issues, but at least nowadays, their sequential aspect should no longer be exposed.

      So that’s one more reason to be careful when exposing DB id in APIs, even if converted to a natural looking key or at least something easier to remember.

    •  Ephera   ( @Ephera@lemmy.ml ) 
      3•3 months ago

      There can also be circumstances where you have to offer people a natural-looking key for general consumption. You can’t put UUID’s on car plates for example.

      Often times, the first section of the UUID is unique enough. With certain UI design choices, one can encourage users to normally work with that, while having the full UUID available in a detail view or from a copy-button.

      Another strategy I quite like, is to have the UUID as the definitely-always-unique identifier, and then have a separate name, which either the users can enter or we generate something like random adjective+animal.

      But yes, neither of those strategies would work for car plates.

      •  lemmyvore   ( @lemmyvore@feddit.nl ) English
        4•3 months ago

        Speaking of car plates, the Wikipedia pages for “Vehicle license plates of [insert country here]” are a rabbit hole.

        I was just reading the page for Romania the other day, speaking of uniqueness, and they had this issue apparently where the combinations overall were enough for the whole country but not enough for their capital city, so they had to hack an extra digit into the plates for the capital.

      •  Skydancer   ( @Skydancer@pawb.social ) 
        12•3 months ago

        There’s nothing wrong with the example in and of itself, but the word “transsexual” in place of “transgender” is not generally random. It is explicitly chosen by Trans-Exclusionary Radical Feminists (TERFs) as well as by right-wing transphobes as a dog whistle to conflate gender dysphoria with drag queens and cross-dess fetishists so as to delegitimise transpeople and suggest some sort of sexual deviance. Coupled with the equivocation of “perceived” gender, motive doesn’t even have to come into it. The words themselves and the concepts they reinforce are transphobic and harmful.

        A witch hunt would have been for me to say that the author is a transphobic asshole whose writings need to be wiped from the internet - which is very far from what I actually posted, which was regret for the way the language they chose distracted from the flow of their argument by reinforcing the social stigmatization of trans people. (Edit: That was a deliberate choice on my part. Not knowing enough about the author to be sure of motives and having no desire to deep dive into their history, I decided that it was only appropriate to point out the hurtful nature of the language and not imply motive.)

        •  AwesomeLowlander   ( @awesome_lowlander@lemmy.dbzer0.com ) 
          9•3 months ago

          It is explicitly chosen by Trans-Exclusionary Radical Feminists (TERFs) as well as by right-wing transphobes

          It can also be used by people with no agenda, including most of the non-western world. Language policing is ridiculous. You want to cancel the transphobes? Stop giving them power and reclaim their terms from them.

        •  TehPers   ( @TehPers@beehaw.org ) English
          5•3 months ago

          Speaking as someone with a MTF close friend and NB spouse, but the term used in the article is the term everyone around me used when I was growing up. That term may be obsolete now, and if so, the author simply needs to be informed. There’s no need to assume they meant harm by it.

          If they knowingly used a term that may offend, then that’s of course a separate issue.

    •  Redkey   ( @Redkey@programming.dev ) 
      7•3 months ago

      I think I can appreciate where you’re coming from, but in the context of the article it was legitimately necessary to address the topic somehow; it’s not like it was written apropos of nothing as a commentary on transsexuality. As a CIS person, I also have a “percieved gender” with which I identify.

      Would “post-transition gender” be a more sensitive term, or less?

      •  Skydancer   ( @Skydancer@pawb.social ) 
        7•3 months ago

        More, but there’s an even simpler solution. In the context, the author is distinguishing between “sex assigned at birth” and “perceived gender.” The equivocating word " perceived" could simply be dropped with no loss of clarity.

  •  kevincox   ( @kevincox@lemmy.ml ) 
    12•3 months ago

    It is true, don’t do it.

    Even at huge companies like Google, lots of stuff was keyed on your email address. This was a huge problem so Google employees were not allowed to change their email for the longest time. Eventually they opened it up by request but they made it very clear that you would run into problems. So many systems and services would break. Over time I think most external services are pretty robust now, but lots of internal systems still use emails (or the username part of it) and have issues.

    IIUC Google accounts now use a random number as the key. But there are still places where the email is in use, slowly being fixed at massive cost.