• Bill (unregistered)

    I would have been first – but I'm in the wrong time zone

  • Kasper (unregistered)

    If you had one dedicated build machine, which would periodically pull from version control and build, this issue would never have happened. Additionally, it gives constant checks that you can actually build what is in version control. Imagine what local changes might otherwise exist on developer's machines, which could lead to differences in the build result.

  • Jakob (unregistered) in reply to Kasper
    Kasper:
    If you had one dedicated build machine, which would periodically pull from version control and build, this issue would never have happened. Additionally, it gives constant checks that you can actually build what is in version control. Imagine what local changes might otherwise exist on developer's machines, which could lead to differences in the build result.

    This sounds like a job for something like, I dunno, BuildMaster maybe?

  • (cs)

    Mmmmm.... unicode lunch wait. *drools

    Also, I wouldn't want to sail on a textile ship...

  • (cs)

    Wouldn't it be 2.11.20120102094553 for Brisbane? Still a WTF, no one works during the 2 week Christmas break: everyone's on summer holidays!

  • (cs)

    TRWTF is IE.

  • ANON (unregistered)

    TRWTF is time

  • (cs) in reply to Zacrath
    Zacrath:
    TRWTF is IE.
    No kidding. I ran into that same problem in a wiki. I came up with a clever transparent div-over-image to allow links to an article. But it doesn't work in IE because transparent divs do not receive the click; it falls through to whatever is not transparent behind behind the div.
  • (cs) in reply to ANON
    ANON:
    TRWTF is time

    TRWTF is time zones.

  • William (unregistered) in reply to Coyne
    Coyne:
    Zacrath:
    TRWTF is IE.
    No kidding. I ran into that same problem in a wiki. I came up with a clever transparent div-over-image to allow links to an article. But it doesn't work in IE because transparent divs do not receive the click; it falls through to whatever is not transparent behind behind the div.

    Easy - just use opacity: 0 instead of background-color: transparent. I use this to style file input controls.

  • RFox (unregistered)

    Never would have happened if the build versions would have used LaTex or TeX's versioning scheme.

    But then TRWTF would be LaTex and TeX

  • faoileag (unregistered)
    Sally:
    We turned off one of our regression tests by accident before launch. The CSS menus don't even appear when you click on a menu item.
    This is probably just improper embellishment (looking at the author, not exactly unlikely), but still. I would expect regression tests to be done by human testers, and those human testers to be self-operable to a degree that you can't switch off a test accidentally.

    You can switch off a test in automated testing, but I don't think they are doing that. Automated testing is usually done on a regular basis, not just before launch.

    As for Dave, who seems to be the one responsible for the menues, he should not have given his ok for rollout before having assured himself that the changes work on all relevant browsers.

    IE7 and IE8 are known problem cases to such a degree that I would even expect the developers themselves to have an occasional glimpse on how their stuff works on IE7 and IE8 while they are still working on it. IETester is a good tool for that.

  • Nobulate (unregistered)

    TRWTF is having accolade meetings.

  • (cs) in reply to faoileag
    faoileag:
    ... I would expect regression tests to be done by human testers, and those human testers to be self-operable to a degree that you can't switch off a test accidentally. ...

    What Sally actually meant, was:

    We forgot to brew a fresh pot of coffee before lunch, so the regression testers all fell into a caffeine-deprived coma

    At least, if my experience of regresison testing is anything to go by... ;)

  • Wrexham (unregistered)
    Pavel groaned. "Of course, mate." He muttered something that Dave could never repeat and typed for a few minutes. "It's on build 2.11.20120101094553."

    094553? I thought the last build was 194553. "Thanks, Pavel."

    Dave hurried to Sally's desk. "Do you have 194553?"

    "One moment." Sally looked through the repo. "No, no such build."

    "Well, that's what's installed in Australia."

    Pavel says that the build installed in Australia is 094553, then Dave tells Sally that 194553 is installed there.

    It looks as though build numbers have got confused in the retelling of the story, as well as in the original happening of the story.

  • agbeladem (unregistered)

    Summary of the WTF:

    A company uses timestamps for their build version numbers, but they forgot to think about timezones.

    Ha ha.

    Omg that's such a hilarious original content!! lmaoroflxd

  • faoileag (unregistered) in reply to agbeladem
    agbeladem:
    Summary of the WTF: A company uses timestamps for their build version numbers, but they forgot to think about timezones.
    You forgot the original WTF: deploying a web application that uses some css wizardry without testing it on IE 8 although major customers still rely heavily on that browser.
  • iWantToKeepAnon (unregistered)
    The Build of the Baskervilles
    The Git is Afoot
    Elementary, My Dear Subversion

    I haven't read the article yet, but the titles already have me chuckling. Nice story telling so far!

  • faoileag (unregistered) in reply to JimM
    JimM:
    What Sally actually meant, was:
    We forgot to brew a fresh pot of coffee before lunch, so the regression testers all fell into a caffeine-deprived coma
    At least, if my experience of regresison testing is anything to go by... ;)
    Of course, such a scenario will soon belong to the past.

    In the day and age of smartwatches that can measure biomedical data, if the coffein-level in a developer reaches dangerously low levels, I expect a notification to be sent off to the nearest Starbucks or <insert_favourite_barista> automatically, who will seen deliver the replenishment right on time.

    "When my fridge starts talking to my grocer, why shouldn't my smartwatch talk to my coffein supplier?"

  • faoileag (unregistered) in reply to iWantToKeepAnon
    iWantToKeepAnon:
    The Build of the Baskervilles
    The Git is Afoot
    Elementary, My Dear Subversion

    I haven't read the article yet, but the titles already have me chuckling. Nice story telling so far!

    Hush, you are only encouraging him!

  • faoileag (unregistered)
    Jason from the build team said:
    Right, well, we append the Unix timestamp at the end of every build version
    Only what they append is not the Unix Timestamp, which is usually considered to be seconds since epoch time.

    Something like "20120101094553" is the format usually found in databases (YYYY-MM-DD HH:MM:SS), regexed into an integer.

  • Nobulate (unregistered) in reply to iWantToKeepAnon
    iWantToKeepAnon:
    The Build of the Baskervilles
    The Git is Afoot
    Elementary, My Dear Subversion

    I haven't read the article yet, but the titles already have me chuckling. Nice story telling so far!

    Hey this is not Slashdot, don't go jumping ahead into comments without RTFA.

  • Nobulate (unregistered) in reply to faoileag
    faoileag:
    "When my fridge starts talking to my grocer, why shouldn't my smartwatch talk to my coffein supplier?"

    coffein is what you get buried in. You probably meant caffeine?

  • MrBester (unregistered) in reply to faoileag
    faoileag:
    IE7 and IE8 are known problem cases to such a degree that I would even expect the developers themselves to have an occasional glimpse on how their stuff works on IE7 and IE8 while they are still working on it. IETester is a good tool for that.

    Bzzzt Wrong. The only way to properly test IEx is to have a native installation. All the kludges like IETester, MultiIE and dicking about in the Dev Tools to try and fake a lower version just don't work properly.

  • QJo (unregistered) in reply to faoileag
    faoileag:
    JimM:
    What Sally actually meant, was:
    We forgot to brew a fresh pot of coffee before lunch, so the regression testers all fell into a caffeine-deprived coma
    At least, if my experience of regresison testing is anything to go by... ;)
    Of course, such a scenario will soon belong to the past.

    In the day and age of smartwatches that can measure biomedical data, if the coffein-level in a developer reaches dangerously low levels, I expect a notification to be sent off to the nearest Starbucks or <insert_favourite_barista> automatically, who will seen deliver the replenishment right on time.

    "When my fridge starts talking to my grocer, why shouldn't my smartwatch talk to my coffein supplier?"

    The real WTF is paying exorbitant amounts to a company that successfully avoids paying the tax they morally ought to pay for average-quality coffee in an unpleasant container just because it's fashionable to be served by a "barista".

  • faoileag (unregistered) in reply to Nobulate
    Nobulate:
    faoileag:
    "When my fridge starts talking to my grocer, why shouldn't my smartwatch talk to my coffein supplier?"

    coffein is what you get buried in. You probably meant caffeine?

    You are right. I think I'll need some more caffeine...

  • faoileag (unregistered) in reply to QJo
    QJo:
    The real WTF is paying exorbitant amounts to a company that successfully avoids paying the tax they morally ought to pay for average-quality coffee in an unpleasant container just because it's fashionable to be served by a "barista".
    Well, not everyone has one of these close enough to his or her place of work .
  • faoileag (unregistered) in reply to MrBester
    MrBester:
    faoileag:
    IE7 and IE8 are known problem cases ... IETester is a good tool for that.

    Bzzzt Wrong. The only way to properly test IEx is to have a native installation. All the kludges like IETester, MultiIE and dicking about in the Dev Tools to try and fake a lower version just don't work properly.

    In my experience they do work properly, unless you need alerts() in Javascript.

    At least they do work well enough so that the problems with the menues mentioned in the article would have been detected far earlier in the process.

    They might give you false positives, however, so getting a VM with the real thing is definitely better, not to mention regression testing on a real machine with XP and and IE 8.

    But for the development process it is definitely better to test on IETester (as IE8) than not to test at all.

  • QJo (unregistered) in reply to faoileag
    faoileag:
    QJo:
    The real WTF is paying exorbitant amounts to a company that successfully avoids paying the tax they morally ought to pay for average-quality coffee in an unpleasant container just because it's fashionable to be served by a "barista".
    Well, not everyone has one of these close enough to his or her place of work .

    The [i]real[/i> real WTF is working for companies who do not make available a supply of free filter coffee (and instant coffee for those who prefer it) which may be consumed in as extravagant quantities as desired.

  • (cs) in reply to faoileag
    faoileag:
    They might give you false positives, however, so getting a VM with the real thing is definitely better

    Luckily, it's easy to do so by using this site.

    But I do agree, even if it should be obvious, some testing is better than no testing. We shouldn't really Nirvana Fallacy the tests - "Oh, this test would not cover 100% of the cases, so we won't use it."

  • ¯\(°_o)/¯ I DUNNO LOL (unregistered) in reply to DaveK
    DaveK:
    Mmmmm.... unicode lunch wait.
    Welcome to The Daily UTF WTF!
  • Leo (unregistered)

    Developers introduce a small bug; when confronted they say "oh yeah" and fix it.

    I guess they can't all be Paula Bean brillantly photographed on a wooden table, but this seems awfully minor for a "WTF".

  • faoileag (unregistered) in reply to Leo
    Leo:
    Developers introduce a small bug; when confronted they say "oh yeah" and fix it. ... this seems awfully minor for a "WTF".
    You call having to roll back the launch of a new version of a web application because after the launch it doesn't work in the browsers of "a dozen high-priority clients" any more "a minor wtf"?
  • Paul Neumann (unregistered) in reply to faoileag
    faoileag:
    You call having to roll back the launch of a new version of a web application because after the launch it doesn't work in the browsers of "a dozen high-priority clients" any more "a minor wtf"?

    Yes. If you don't, never ever ever leave your current employ.

  • (cs) in reply to faoileag
    faoileag:
    You call having to roll back the launch of a new version of a web application because after the launch it doesn't work in the browsers of "a dozen high-priority clients" any more "a minor wtf"?

    tbh I think the biggest WTF is their deployment protocol - afaict they're pushing source code packages out as patches, and having clients compile the code locally, except when it's "their" office in Boston, which ... erm, nope, I'm lost. What's the protocal again? Doesn't sound much like a web app to me.....?

  • (cs) in reply to faoileag
    faoileag:
    Jason from the build team said:
    Right, well, we append the Unix timestamp at the end of every build version
    Only what they append is not the Unix Timestamp, which is usually considered to be seconds since epoch time.
    ... and which does not depend on timezone (epoch is 1970-01-01 00:00 UTC).
  • faoileag (unregistered) in reply to Paul Neumann
    Paul Neumann:
    faoileag:
    You call having to roll back the launch of a new version of a web application because after the launch it doesn't work in the browsers of "a dozen high-priority clients" any more "a minor wtf"?
    Yes. If you don't, never ever ever leave your current employ.
    If a customer ui that isn't working any more for the company's high priority clients would be considered a minor wtf in my current employ, I would probably start updating my cv and looking elsewhere.

    It depends on the situation, but a broken ui can mean lost revenue summing up by the minute (think transaction fees for an online brokerage), unused resources at a factory (if the truck driver can't change the time window at the loading bay) or your offspring loosing a place at his or her favourite college because registration broke down.

  • (cs)

    This is the first good article I've seen by Erik, complete with an actual WTF and lack of a snarky pseudo-samurai with a case of unwarranted self importance. Keep up the good work.

  • (cs) in reply to William
    William:
    Coyne:
    Zacrath:
    TRWTF is IE.
    No kidding. I ran into that same problem in a wiki. I came up with a clever transparent div-over-image to allow links to an article. But it doesn't work in IE because transparent divs do not receive the click; it falls through to whatever is not transparent behind behind the div.

    Easy - just use opacity: 0 instead of background-color: transparent. I use this to style file input controls.

    That's a very useful hint. Thank you.

    But it doesn't change the fact that IE does "background-color: transparent" different than Firefox and Chrome and even Safari. Which IMHO makes IE still TRWTF.

  • n_slash_a (unregistered) in reply to faoileag
    faoileag:
    agbeladem:
    Summary of the WTF: A company uses timestamps for their build version numbers, but they forgot to think about timezones.
    You forgot the original WTF: deploying a web application that uses some css wizardry without testing it on IE 8 although major customers still rely heavily on that browser.
    If only, we just recently (last year) were able move off of IE6. IE8 is a dream by comparison. Of course, everyone I know uses Firefox or Chrome anyway :)
  • (cs) in reply to faoileag
    faoileag:
    … or your offspring loosing a place at his or her favourite college because registration broke down.
    If it is a good one, the college is probably fairly OK with that; they'll have plenty of other people queueing up to attend who don't use old versions of IE.
  • J (unregistered)

    You go to meetings where you are commended for your achievements? Seriously, dude, count your blessings.

  • Developer Dude (unregistered) in reply to Zacrath
    Zacrath:
    TRWTF is IE.

    Quite common actually.

    A lot of big orgs/corps are a years behind in many respects, sometimes as much as a decade. I know of one VERY large corp that until very recently would install WinXP on all machines that came in and only approve IE6 as a browser - not allowing any other browser to be installed (naughty naughty).

    The real WTF was when they came out with a notice that IE6 was insecure for surfing the net (no big surprise) and recommended IE6 only for internal websites and recommended using a different browser for external websites while apparently never lifting the restriction on installing any browser except IE6.

    WTF?

  • Klimax (unregistered)

    Funnily enough, IE supported many effects for longer... http://msdn.microsoft.com/en-us/library/ie/ms532853(v=vs.85).aspx

    (Alongside of VML it is possible to create quite complex things, which with help of supporting JS libraries - placing either SVG+CSS3 or MS-filters+VML - can work with IE6+ alongside of newer browsers)

    So in these cases like transparency, it shouldn't be that hard to support even quite old version of IE.

  • DQ (unregistered) in reply to Developer Dude
    Developer Dude:
    Zacrath:
    TRWTF is IE.

    Quite common actually.

    A lot of big orgs/corps are a years behind in many respects, sometimes as much as a decade. I know of one VERY large corp that until very recently would install WinXP on all machines that came in and only approve IE6 as a browser - not allowing any other browser to be installed (naughty naughty).

    The real WTF was when they came out with a notice that IE6 was insecure for surfing the net (no big surprise) and recommended IE6 only for internal websites and recommended using a different browser for external websites while apparently never lifting the restriction on installing any browser except IE6.

    WTF?

    We're still on Windows XP, but at least we have IE8 :) And it's only since last month we upgraded from Office 2003 to 2010. Yeah, only 3 years behind now. Oh, and soon they will start testing if the upgrade doesn't cause any issues. Nice upgrade planning!

  • C-Derb (unregistered) in reply to Klimax
    Klimax:
    Funnily enough, IE supported many effects for longer... http://msdn.microsoft.com/en-us/library/ie/ms532853(v=vs.85).aspx

    (Alongside of VML it is possible to create quite complex things, which with help of supporting JS libraries - placing either SVG+CSS3 or MS-filters+VML - can work with IE6+ alongside of newer browsers)

    So in these cases like transparency, it shouldn't be that hard to support even quite old version of IE.

    The question isn't whether old versions of IE can be supported, but rather should they be supported. I'm sure there's a few clients somewhere that are big enough to cater to, but most clients should be told to bugger off if they insist on using outdated IE versions. The ones who insist on continued support for a horrendous and outdated browser should be charged a premium because it just isn't cost effective to support the shit that is IE 6 or IE7.

    Disclaimer: I hate IE.

  • (cs) in reply to Dogsworth
    Dogsworth:
    This is the first good article I've seen by Erik, complete with an actual WTF and lack of a snarky pseudo-samurai with a case of unwarranted self importance. Keep up the good work.

    +1

    It's actually the first article that I've seen in a long time where the stuff that you can tell was added to the original story actually added the enjoyment of reading it.

  • (cs) in reply to Developer Dude
    Developer Dude:
    Zacrath:
    TRWTF is IE.

    Quite common actually.

    A lot of big orgs/corps are a years behind in many respects, sometimes as much as a decade. I know of one VERY large corp that until very recently would install WinXP on all machines that came in and only approve IE6 as a browser - not allowing any other browser to be installed (naughty naughty).

    The real WTF was when they came out with a notice that IE6 was insecure for surfing the net (no big surprise) and recommended IE6 only for internal websites and recommended using a different browser for external websites while apparently never lifting the restriction on installing any browser except IE6.

    WTF?

    Not that bad. I had a client that forced its users to only have Netscape Communicator 4. In 2007.

  • anonymous (unregistered)

    Wait, I'm confused. What did the fact that they broke the build have to do with the fact that their versioning system was fucked, other than the fact that the need to roll back the build led them to discover it?

  • Sarcasto (unregistered) in reply to QJo
    QJo:
    faoileag:
    QJo:
    The real WTF is paying exorbitant amounts to a company that successfully avoids paying the tax they morally ought to pay for average-quality coffee in an unpleasant container just because it's fashionable to be served by a "barista".
    Well, not everyone has one of these close enough to his or her place of work .

    The [i]real[/i> real WTF is working for companies who do not make available a supply of free filter coffee (and instant coffee for those who prefer it) which may be consumed in as extravagant quantities as desired.

    Well, if you drink tea the W3c has you covered with HTTP 418:

    http://en.wikipedia.org/wiki/List_of_HTTP_status_codes

Leave a comment on “The Build of the Baskervilles”

Log In or post as a guest

Replying to comment #:

« Return to Article