Discourse DMZ



  • @eviltrout said:

    You thought wrong.

    And yet somehow there's still 36421434 bugs and we find new ones every fucking day.

    Whatever you're doing, it doesn't work.



  • Then I am really sad for discourse, because the amount of UI regressions is astronomical. Your tests are equivalent to assert(true) and should be significantly rethought.

    Let's start with something simple:

    1a) My avatar (blank) linked to meta.discourse.org (my avatar over there), and instead of showing my avatar from there, it was blank.
    1b) Linking the same image in my about me section worked just fine
    1c) Discourse eventually gave up and blanked out my avatar entirely, leaving it a transparent 16x16 image
    1d) This image isn't actually transparent on mobile, and instead leaves a white box in the blue header, but on desktop there is no white box in the header.

    2a) Everything about this post:
    http://what.thedailywtf.com/t/dashes-a-k-a-blakey-needs-to-fix-contrast-on-his-monitor/2922/56?u=matches

    3a) Everything about that quote from 2a


  • Banned

    We're not perfect, but I think we do a lot right and are getting better. It's cool if you guys disagree but we are trying our best.



  • I think the whole "We're ready for 1.0!" thing was a lot of people's problem.

    The whole thing is still packed with bugs, trying your best would be a lot more appreciated if this wasn't being touted as a ready to go product...


  • I survived the hour long Uno hand

    I was considering offering some suggestions but apparently ya'll don't want my help.


  • Banned

    @KillaCoder said:

    I think the whole "We're ready for 1.0!" thing was a lot of people's problem.

    I think a lot people here were hating on Discourse way before that ever happened.

    Obviously everyone has different standards of quality, and I appreciate all the bug reports you guys have been doing, but you have to realize there is a serious difference of perspective between here and virtually every other discourse install in terms of "ermaghered it's overflowing with bugs."



  • Since it's been installed here, we've heard literally every day 'IT'S ALMOST 1.0 GUYS! HOME STRETCH!'

    Fine, but your 1.0 is a mass of bugs. And that's cool, but the emphasis you put on it is '1.0 is gold, we're so good, lets change heatmaps and leader to veteran! and other trivial changes that have NOTHING to do with bug fixes!'

    If instead it's been pitched as 'Hey, it's 1.0. It's not great, but in general it works the way we want it to and is the start of something great. It has issues, but we're working on them'

    Expectations would be very different, and you'd have a much different reaction to the software.

    1.0 is not gold. 1.0 is still .9.8.3.whateverthefuckitwas and .9.8.3 is really closer to .3 or .4 of an actual release. If you sold this on steam, people would be pissed at the garbage issues and regressions we find literally every day. Our Bug category has more posts than our meta category.


  • I survived the hour long Uno hand

    @Matches said:

    Our Bug category has more posts than our meta category.

    We should sign up for a free account on a bug tracker, stick all the bugs we find in it, and toss it at the meta. Like, "we can't make you fix shit, but at least we can keep an accurate tally without deleting or losing reports".



  • Only if you build the integration so that I can post in the bugs category and it gets shipped off there, because unless they pay me, I'm not spending more time than it takes to post the bug.

    I've spent a lot of time finding bugs in the software with repro steps. I was reporting them to meta.d as well, until Jeff decided it's better to just delete my reports rather than address them.

    Honestly, if Jeff was removed from the project, and they started using an actual bug tracker, with actual targets on when things would be fixed I would switch my stance from 'vote of no confidence' to 'Could work, pending the core team'


  • Banned

    @Matches said:

    If you sold this on steam, people would be pissed at the garbage issues and regressions

    Well the thing is we do sell Discourse, have a bunch of customers and literally nobody is complaining about the bugs like you guys are. Even from the many forums out there that aren't paying us we aren't seeing this. I know it's hard to believe but it's the truth.

    I'm not saying there aren't bugs, there are. It's just the whole chicken little "oh my god the sky is falling" is not something we hear from LITERALLY ANYONE ELSE.

    But hey, what else should I expect from a community that's entirely based on making fun of other people's code?



  • @eviltrout said:

    We're not perfect, but I think we do a lot right and are getting better. It's cool if you guys disagree but we are trying our best.

    You're honestly saying that me typing, "3." and seeing "1." on the screen is trying your best?

    You're honestly saying that resolving bugs with, "oh one person on the Discourse team personally doesn't use that page, so CLOSED!!!" is trying your best?

    If this is trying your best, you are not very good at software development. Sorry but it's true.

    @eviltrout said:

    I think a lot people here were hating on Discourse way before that ever happened.

    You're wrong on this point. At the time Discourse was deployed here, it was only days from 1.0, according to your own lead developer. It was like version 0.9999989999991 or something.

    We (well, except Ben L.) never experienced what you guys considered "beta", we only experienced, "ready for 1.0!!!"

    @eviltrout said:

    Obviously everyone has different standards of quality,

    Right; the really discouraging part is you, the developer of the product are the one serving up this shit and saying, "isn't this delicious!" The developer should have a much stronger sense of quality than the user. You don't. You're apparently perfectly ok with this shit sandwich.

    And in any case, if you honestly believe this is an acceptable level of quality, we're right back to: you are not very good at software development. Sorry but it's true.

    @eviltrout said:

    and I appreciate all the bug reports you guys have been doing,

    You might, but Atwood certainly doesn't. And since he controls (with Nazi-like passion) your "bug tracker" forum, we get the distinct impression you do not give a shit.

    Oh, also, the fact that none of our reported bugs ever get fixed also kind of contributes to the "you do not give a shit" impression we got going here.

    @eviltrout said:

    but you have to realize there is a serious difference of perspective between here and virtually every other discourse install in terms of "ermaghered it's overflowing with bugs."

    I don't give a shit what they think. I'm here because I like the users here, not because of the software. Your software is holding me hostage. I'm FORCED to use it to interact with the community here. Do you understand that point? Because Atwood certainly doesn't.

    I don't have the choice to just use Community Server, or just use PHPBB, or just use something that meets my needs more easily. Believe me, if I could-- you guys even for a single millisecond threw me a fucking bone and gave me a normal paginated interface-- you bet your ass I'd be using it.

    I don't give a shit what Atwood or you think a "good" forum looks like. Do you understand that? As long as I'm forced to use his vision, and as long as he stubbornly refuses to yield to anybody else's conception of what forum software should look like, I'm probably never going to be happy with Discourse.

    That said, the fact that it's a buggy mess certainly isn't helping the situation.



  • Lets try something else, a simple concept.

    Last activity coldmap indicates it's an ice cold topic, but the last activity is 34m / 1 hour ago.

    The likes content indicates it's a hot topic, with lots of love going on.

    These things DO NOT AGREE with each other.



  • @Matches said:

    1.0 is not gold. 1.0 is still .9.8.3.whateverthefuckitwas and .9.8.3 is really closer to .3 or .4 of an actual release. If you sold this on steam, people would be pissed at the garbage issues and regressions we find literally every day. Our Bug category has more posts than our meta category.

    Discourse is classic CADT development. Version 0.8 is always followed by version 0.8 which is always followed by version 0.8.


  • Banned

    @blakeyrat said:

    I don't give a shit what they think. I'm here because I like the users here, not because of the software. Your software is holding me hostage. I'm FORCED to use it to interact with the community here. Do you understand that point?

    Cool, I'll just put your special snowflake needs above everyone else's, because that would make me a Good Software Developer(tm).

    I'm done replying here BTW, it's taking up too much of my day!



  • @eviltrout said:

    Well the thing is we do sell Discourse,

    Right, and I hope you feel guilty for cheating all those people with this crap.

    @eviltrout said:

    have a bunch of customers and literally nobody is complaining about the bugs like you guys are.

    This isn't relevant. We're not talking about other people. We're talking about two things:

    1. The people here who are forced to use your product, and
    2. Your own sense of professionalism, and sense for what is acceptable quality

    @eviltrout said:

    I'm not saying there aren't bugs, there are.

    Then stop changing pointless shit like the "heatmap" (or whatever the fuck Jeff thinks a "heatmap" is) and start fixing bugs.

    Here's a good way to speak in actions instead of words: how about you take that simple bug where typing a "3." in the text box results in a "1." in the post, and you re-open that bug report, and you acknowledge that it is a valid bug, and you fucking fix it?

    @eviltrout said:

    But hey, what else should I expect from a community that's entirely based on making fun of other people's code?

    We're not the problem here. We've given you thousands or tens of thousands of hours of free testing services. And you repay us by dismissing us with that bullshit?

    How about you cut everybody in our community who's fixed your goddamned bugs a fucking check? Maybe some actions instead of words for once?



  • @eviltrout said:

    Cool, I'll just put your special snowflake needs above everyone else's, because that would make me a Good Software Developer(tm).

    Translation: "No, I do not understand your point."



  • @blakeyrat said:

    And yet somehow there's still 36421434 bugs and we find new ones every fucking day

    Not entirely true. Sometimes we're just finding regressions. But probably close enough for this discussion.


  • ♿ (Parody)

    @blakeyrat said:

    What strawman, I'm looking at the end-product of this testing philosophy right now. With my two eyes. It's on the computer monitor I'm looking at. While I'm typing this.

    No. No one said that only automated testing should be done. But any time someone mentions any sort of automated testing, you begin flaming that only doing automated testing is wrong. Which no one advocated.



  • The number is also usually between 1-3, but yeah.



  • @eviltrout said:

    I think a lot people here were hating on Discourse way before that ever happened.

    And that had more to do with Jeff's attitude than anything.


  • ♿ (Parody)

    @eviltrout said:

    It's cool if you guys disagree but we are trying our best.

    I don't doubt it. I've looked at a few commits here and there, and for the most part, they don't include any changes to any tests. I guess for CSS tweaks, that may be reasonable, but especially now that you guys are claiming v1.0, requiring (within reason) tests for features and fixes seems like a good idea.


  • ♿ (Parody)

    @Matches said:

    Lets try something else, a simple concept.

    Last activity coldmap indicates it's an ice cold topic, but the last activity is 34m / 1 hour ago.

    The likes content indicates it's a hot topic, with lots of love going on.

    These things DO NOT AGREE with each other.

    Wrong. This is an example of MDGI.



  • @Yamikuronue said:

    it only makes sense to do automated smoke tests to ensure something totally broken is never deployed.

    They've fucking failed on that from day 1



  • @boomzilla said:

    No. No one said that only automated testing should be done.

    Every team I've worked on or been exposed to that did automated testing only did automated testing. EDIT: I actually interviewed at a company a few months ago that was so proud that they got rid of their QA department. I think I lost the job because I spent a lot of time talking about the benefits of having people dedicated to QA before they dropped that bomb on me.

    Maybe I've only worked for weird atypical companies. I doubt it, but I concede the possibility.



  • Huh. My post just vanished, I think TDWTF is about to have a heart attack again.

    MDGI?
    MDGI?
    MDGI?
    MDGI?
    MDGI?
    MDGI?



  • Oh, hey, look, a regression:



  • That was well played.


  • ♿ (Parody)

    @blakeyrat said:

    Every team I've worked on or been exposed to that did automated testing only did automated testing. EDIT: I actually interviewed at a company a few months ago that was so proud that they got rid of their QA department.

    Amazing. We have both where I am. I just caught a regression yesterday from automated testing. Sadly, our human testers find a lot more.



  • Generally automated testing follows an 80-20 rule, 80% of the stuff needing to be tested can be done so using machines, and for the most part it's great wonderful and perfect (assuming decent tests).

    20% comes from looking at new features, making sure tests behave as expected, make sure something is not 'technically right' but 'humanly wrong' [working as designed but not as expected] and a slew of other things.

    You can't replace your QA department, but 3-5 good testers should catch damn near everything (at least on dedicated software, if you manage a lot of products quality is likely to suffer with only 3-5 people)

    QA in general should have no less than 2 people testing the software, and those people should ideally not be the same people developing the software. (Not always possible in very small shops (only 2-3 people in the entire company)/indie, but absolutely possible for normal small-medium-large businesses)


  • I survived the hour long Uno hand

    Capers Jones found that you want at least 3% QA staff. So if your staff is under 100, 3-5 is plenty.



  • The amount of QA people you need is entirely dependent on

    1. Number of products you support
    2. Number of new features you push out
    3. How much 'human interface' you have (a product that is 100% api text needs significantly less human interaction than a product that is 80% UI)
    4. How critical your software is
      4b. How willing you are to accept downtime / issues
      4c. If human risk to life is possible
    5. Many other factors that I'm not going to list here because fuck.

  • BINNED

    @blakeyrat said:

    I'm looking at the end-product of this testing philosophy right now.

    Which testing philosophy?

    • the one that says testing can be 100% automated, or
    • the one that says testing isn't going to find all bugs so just ship it, or
    • "It's working to my satisfaction so it's all good" (be careful with this one because glass houses, etc.)

    I only remember seeing the second and third from Atwood.



  • I assume he means whatever discourse testing philosophy is.



  • @Matches said:

    There are some testing suites to help with UI testing,

    Watir (in Ruby) and watin (C#) come to mind.

    I can only speak for watin and that's rather slow. Too slow for per-release testing if you want to release more than once a day.

    We let it run in the nightly builds where I worked a couple of years ago, and it was ok, although not 100% reliable.



  • @blakeyrat Totally get what you're saying. I had a similar experience recently where I typed "2." and got a "1." and it literally ruined my whole week!


  • ♿ (Parody)

    @Matches said:

    I assume he means whatever he thinks discourse testing philosophy is.

    FTFY

    But what does he think the philosophy is?


  • ♿ (Parody)

    @faoileag said:

    Watir (in Ruby) and watin (C#) come to mind.

    There's also selenium, which drives a web browser and tests stuff.



  • So @eviltrout, the nicest Discourse developer, comes over here, and we berate him. Awesome.



  • I nominate @sam as the nicest.


  • Winner of the 2016 Presidential Election

    I liked @riking, too!


  • ♿ (Parody)

    @chubertdev said:

    So @eviltrout, the nicest Discourse developer, comes over here, and we berate him. Awesome.

    Only because we can't berate him over there or we get deleted.



  • @Matches said:

    I nominate @sam as the nicest.

    Ehhh, he's nice, I guess.

    @Kuro said:

    I liked @riking, too!

    I don't see @riking post enough to have an opinion.


  • ♿ (Parody)

    @chubertdev said:

    I don't see @riking post enough to have an opinion.

    @riking isn't dismissive at all IME. @sam has fought back a bit, but he's generally constructive and willing to talk it out. And when he has an opinion (which I do not always agree with) he generally has decent reasons for them.


  • Winner of the 2016 Presidential Election

    Well, he did fix a bug by abusing said bug to become an admin, give himself a useless badge and fix the bug. Afterwards he openly stood by what he did.

    That alone gives him at least 85% in my book!

    Also: Almost no bugs found while writing the last reply:

    Did the Server just have a hickup?

    Filed Under: I really have very little problem with the Discourse devs over here. Yes, Jeff was a bit difficult for most people here back in the days but really, they are just doing their job. And I at least hope they are having a little fun ;P



  • Yes, but he only gave himself the css badge, he should have gone with the xss badge. So that's like -10% right there.


  • Winner of the 2016 Presidential Election

    I thought he was being modest, seeing as only the best people in this comunity get the XSS-badge cough 😆

    Filed Under: Couldn't keep a straight face



  • @eviltrout said:

    Obviously everyone has different standards of quality, .... but you have to realize there is a serious difference of perspective between here and virtually every other discourse install in terms of "ermaghered it's overflowing with bugs."

    Because mostly you have to be a software developer / tester to notice them. After all, if all you want to do is post a simple sentence with a bit of bold text, or a link, you can do that.

    But like the "normal" Windows XP user won't know if his machine is pwned or not, the "normal" Discourse user won't notice that private email addresses are leaked with some JSON requests.

    However, that the normal user does not notice the bugs doesn't make the software less buggy.

    That's not "different standards of quality". That's the Dunning-Kruger syndrome in action.

    If you take the stand that "no other forum is complaining that much" as a sign of the maturity of Discourse, you are cheating yourself.

    In much the same way that you would cheat yourself if you believed your web server was safe even though the company testing the security didn't do any penetration testing.


  • :belt_onion:

    @Matches said:

    Last activity coldmap indicates it's an ice old topic

    FTFY



  • @blakeyrat, I figured it out.

    They were posting about the version of Discourse.

    It was supposed to be Version 0.000000001.

    Then they started the line with the Version number.

    At Discoursce changed it to Version 1.


  • ♿ (Parody)

    @chubertdev said:

    @blakeyrat, I figured it out.

    They were posting about the version of Discourse.

    It was supposed to be Version 0.000000001.

    Then they started the line with the Version number.

    At Discoursce changed it to Version 1.

    +�


Log in to reply