gsuberland, (edited )
@gsuberland@chaos.social avatar

if you were sorting a parameter by ascending numeric value (in some sort of parameteric search engine) and the values included positive, negative, and ± values all together, where would you want the ± ones to be placed?

context is electronics stuff, where you're searching for a part that fits your required specs.

(boosts welcome for more input!)

mmu_man,
@mmu_man@m.g3l.org avatar

@gsuberland I don't know, just anything but the RS way 😄

joxean,
@joxean@mastodon.social avatar
AMS,

@gsuberland Usually I'd say either split in two (min and max) or give me a matcher than can deal (TI does pretty well).

jaseg,
@jaseg@chaos.social avatar

@gsuberland If it's a supply voltage spec, I'd want +/-12V next to 24V because they are the same. If it's s differential input voltage spec, I'd want it in a separate column because I will likely be searching either for differential parts or for single-ended parts, not for both at the same time. If it's a voltage interval that doesn't necessarily have to be symmetric, I'd want it to be split in two columns, min and max.

gsuberland,
@gsuberland@chaos.social avatar

@jaseg one style for all specs, pick the least bad option.

anymouse_404,
@anymouse_404@glitch.social avatar

@gsuberland If we're talking about things like op-amps, imo they should be sorted and displayed by the absolute value, eg. 5V, 9V, 10V(±5V), 12V, ...
If we are talking about things like power supplies or powered devices, those should be separated. Whether dual supply first or last is question of which one is the "normal" one? Eg. I'd say you'd (traditionally) expect power amplifiers to be symmetric and single rail ones to be more of a special case, but the other way with power supplies.

gsuberland,
@gsuberland@chaos.social avatar

@anymouse_404 any generic spec, not one part.

darthnull,

@gsuberland @xabean I kind of feel like you want “normal” numbers first, and only then the special +/- cases.

“Okay, there’s -4 through +3, oh and also +/- 8 and +/- 12.”

The other way (putting them first) throws my brain in “gotta pay real close attention to these odd cases” mode and I don’t think they intuitively register as well.

gsuberland,
@gsuberland@chaos.social avatar

@darthnull @xabean I, too, am somewhat surprised that "last" got such a low result.

zethtren,
@zethtren@mastodon.social avatar

@gsuberland Can we make a T? And treat them like they’re on the imaginary axis?

gsuberland,
@gsuberland@chaos.social avatar

@zethtren only if you can fit an extra axis into a dropdown box somehow ;)

zethtren,
@zethtren@mastodon.social avatar

@gsuberland If the dropdown can then drop sideways 😂

jpm,
@jpm@aus.social avatar

@gsuberland this is absolutely cursed, and something I’ve never considered

gsuberland,
@gsuberland@chaos.social avatar

@jpm turns out dumping part attribute data for electronics stuff and then trying to normalise it in a way that's searchable and sortable is remarkably tricky.

gsuberland,
@gsuberland@chaos.social avatar

@jpm say hello to this wonderful abomination

gsuberland,
@gsuberland@chaos.social avatar

@jpm (and yes, I genuinely considered an ANTLR grammar for this task)

jpm,
@jpm@aus.social avatar

@gsuberland or you could just YOLO everything LCSC-style and end up having LED colours that include 粉灯

gsuberland,
@gsuberland@chaos.social avatar

@jpm this is exactly what I'm aiming to fix :D

xabean,

deleted_by_author

  • Loading...
  • jpm,
    @jpm@aus.social avatar

    @xabean @gsuberland Capacitor: it depends on the chemistry
    Inductor: well…

    xabean,

    deleted_by_author

  • Loading...
  • jpm,
    @jpm@aus.social avatar

    @xabean @gsuberland MOSFET: same, but before I get started…

    gsuberland,
    @gsuberland@chaos.social avatar

    One other option would be "combine with both positive and negative", i.e. duplicate the ± items, but that seems awkward and would be annoying to get right in edge cases (e.g. all values being ± values except a couple of them) so I really don't like it.

    gsuberland, (edited )
    @gsuberland@chaos.social avatar

    next question:

    if there are parameter values that don't fit any standard format when parsed (e.g. unusual multi-element formatting, typos, other outliers) where should they appear when sorted in ascending order?

    "Lexicographically" here means that a weird value like "7AD4P" or "7.38?102" would appear after all the numbers starting with 7, but before the numbers starting with 8.

    consider that this may affect range searches if they are implemented.

    jaseg,
    @jaseg@chaos.social avatar

    @gsuberland I'd want these combined under an "other" value at the beginning of the list, because they're unlikely to be useful as is, and I can just full text search for them when I need them.

    jhwgh1968,
    @jhwgh1968@chaos.social avatar

    deleted_by_author

  • Loading...
  • gsuberland,
    @gsuberland@chaos.social avatar

    @jhwgh1968 "searching for either side includes it in the search" are you referring to value range sliders here?

    jhwgh1968,
    @jhwgh1968@chaos.social avatar

    deleted_by_author

  • Loading...
  • gsuberland,
    @gsuberland@chaos.social avatar

    @jhwgh1968 ah. in this case there's no range search anyway, so that's not a concern.

    I always feel like that functionality needs differing semantics per parameter, too. sometimes -3 and ±3 have distinct meanings and you wouldn't want ±3 inlcuded (e.g. on tolerances)

    jhwgh1968,
    @jhwgh1968@chaos.social avatar

    @gsuberland funny you say that, I almost typed this:

    I presume you mean flexibility in a part, e.g. LM317 and LM337

    as opposed to an inclusive range, e.g. VS and SS rails of an op amp where they always center the middle of the output at zero (and if you want another baseline voltage, you do the math)

    gsuberland,
    @gsuberland@chaos.social avatar

    @jhwgh1968 yeah it ends up being super contextual so it's very difficult to pick a one-size-fits-all range search process for this kind of thing. sometimes you just want a range intersection test, sometimes you want a superset test, sometimes you want a subset test.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • Electronics
  • kavyap
  • thenastyranch
  • cubers
  • ethstaker
  • InstantRegret
  • DreamBathrooms
  • ngwrru68w68
  • magazineikmin
  • everett
  • Youngstown
  • mdbf
  • slotface
  • rosin
  • GTA5RPClips
  • JUstTest
  • khanakhh
  • normalnudes
  • osvaldo12
  • cisconetworking
  • provamag3
  • Durango
  • tacticalgear
  • modclub
  • Leos
  • megavids
  • tester
  • anitta
  • lostlight
  • All magazines