Thread: Regarding getting involved with open-source projects

  1. #1
    Make Fortran great again
    Join Date
    Sep 2009
    Posts
    1,413

    Regarding getting involved with open-source projects

    I think we had a thread sometime recently about the problems with getting into open-source projects. I thought I'd share an experience I had yesterday.

    I have several numerical solvers written in Fortran. I wanted to test them out using quad precision arithmetic. Fortran makes it easy to do this, all you have to do is change the kind parameter of your flaoting-point variables.

    Apparently the quad support is still in its infancy, as some simple calculations caused segfaults on both MinGW and Linux gfortran compilers. So I filed a bug report on the GCC bugtracker.

    Someone from the GCC team apparently saw the word MinGW and basically closed the report off as a MinGW-only problem. When I pointed out that I had tested on GNU/Linux plain GCC, I was met with hostility by this individual and he basically made every attempt to make me look dumb instead of being helpful. Basically made me feel like an .............. when I went out of my way to try and help the project by reporting the bug.

    I had provided a clear description of what the problem was, what compiler flags were responsible, and the code to reproduce the problem. I later posted a GDB backtrace as well.

    I'm not going to link to the bug on the bug tracker, I'll just leave it at my involvement with the bug-fixing process ended as of last night when I called this person another phrase for f-ing female genitals.

    I had heard that it's difficult to get into the GCC and Linux projects. Congrats to the GCC team in holding up their reputation. I hope Clang wins.

  2. #2
    Registered User MutantJohn's Avatar
    Join Date
    Feb 2013
    Posts
    2,665
    Hmm... I googled quadruple precision and it seems like they use 128 bits so that's a long double or a __float128, right? I thought long doubles were stable but I think __float128's are kind of thrown at a wall in hopes they'll stick so I'm not surprised they have issues.

    Kind of lame you were treated like that though

    I'm more than confident, though, that he's not representative of the entire group. I have very rarely met anyone associated with open source projects that is mean. Like, I changed something in Ubuntu and it killed all my desktop/menu icons and I sent this rage email to the Ubuntu team full of swearing and all this stuff and the response was awesome. This person let me know that speaking like that was unacceptable, they were listening to my problems and that if I wanted help, I should be more polite. They were very empathetic with the fact that I'm a user who was told it was safe and it backfired and that would cause some frustration. So I sent this awesome email back and a bunch of different Ubuntu people replied as well. Was very neat.

  3. #3
    Make Fortran great again
    Join Date
    Sep 2009
    Posts
    1,413
    I've been sticking to the GNU libquadmath library, so __float128. I've used libquadmath from C before with no issues. libgfortran's use of libquadmath is either buggy or there's an issue with GCC itself. Basically, the issue is that code using quad precision combined with -march=native on both machines I use causes a segfault. Without the arch option, compilation and execution is fine.

  4. #4
    Make Fortran great again
    Join Date
    Sep 2009
    Posts
    1,413
    I've had much better luck with some projects. It seems like GNU projects are especially bad. I also had a lot of trouble with GNU Octave, which is why I don't help them anymore.

  5. #5
    Ticked and off
    Join Date
    Oct 2011
    Location
    La-la land
    Posts
    1,728
    Don't let one butt-gargler ruin everything. Stagnant academics' first instinct is always to redirect any criticism elsewhere. And GNU projects are full of academics.

    Epy, I can't reproduce your crash, but if you care to send me the objdump -d sip_test output (full disassembly of your binary), and the GDB backtrace and register state at the point of crash, I might be able to help. Compiling the binary static first would help, in case the bug is in some remote corner of the library (say, Fortran bindings) instead of your code.

    I'm debugging the other similar issue (50201) just to .. show that these experiences are common, but having colleagues to help buffer against the abrasive ones helps? .. , but your problem does not look like has the same roots. I shall now go back to my lurking, but you can catch me at GCC bugzilla if you want.

  6. #6
    Make Fortran great again
    Join Date
    Sep 2009
    Posts
    1,413
    That really struck me as stalkerish until you mentioned 50201. Lol. Seriously that guy ........ed me off. I never use gdb, and I've never had anyone ask me for more than simply code required to reproduce a bug. I'm at work right now so I could only give you the Windows MinGW stuff.

  7. #7
    Make Fortran great again
    Join Date
    Sep 2009
    Posts
    1,413
    I managed to reduce my set of files and functions down to this:

    Code:
    SUBROUTINE test(N)
    IMPLICIT NONE
    INTEGER, INTENT(IN) :: N
    REAL(KIND=16) :: array(N)
    array = 0
    END SUBROUTINE test
    
    PROGRAM main
    IMPLICIT NONE
    CALL test(10)
    END PROGRAM main
    Which produces the same segfault at 'array = 0'. Strange, on MinGW compiling with -static eliminates the segfault, but I just tried the same on an Ubuntu server which still segfaulted with -static.

  8. #8
    Unregistered User Yarin's Avatar
    Join Date
    Jul 2007
    Posts
    2,158
    Quote Originally Posted by MutantJohn View Post
    Like, I changed something in Ubuntu and it killed all my desktop/menu icons and I sent this rage email to the Ubuntu team full of swearing and all this stuff and the response was awesome.
    It may be unfair to compare Ubuntu to most FOSS, since it's primarily developed by a for-profit company.

  9. #9
    Make Fortran great again
    Join Date
    Sep 2009
    Posts
    1,413
    John is right though, I've had much better luck with other projects before. GNU projects just seem to be the worst.

  10. #10
    Unregistered User Yarin's Avatar
    Join Date
    Jul 2007
    Posts
    2,158
    Quote Originally Posted by Epy View Post
    GNU projects just seem to be the worst.
    Don't be "negative in the freedom dimension"

  11. #11
    Ticked and off
    Join Date
    Oct 2011
    Location
    La-la land
    Posts
    1,728
    I felt I have to add the following:

    On many projects, the maintainers and sub-maintainers tend to develop a tendency to skim bug reports for keywords, so that they can dismiss the bug report as irrelevant. They do not necessarily even read the report; they may latch on to a single specific word without even reading the entire post, as I believe happened in Epy's case.

    Pointing out the error the dismissive person made, simply causes them to try and reject the bug report or the reporter for other reasons. I suspect the maintainers feel that their prestige as maintainer is threatened if they admit to making an error.

    Compare to for example the attitude Laserlight has shown in this forum. Instead of looking for reasons to dismiss a question, she asks for clarification, and often even gives helpful directions to the asker to how to do so. That's what best maintainers do. Greg Kroah-Hartman at linux-kernel mailing list (LKML) is another good example. LKML otherwise is chock full of well-known abrasive personalities; there preparatory research is necessary to avoid an initial dismissal.

    While one might be tempted to recommend always doing research to find out what triggers these maintainers to avoid them, I advise against that. It takes too much of your time, unless the project is known to have many abrasive personalities. Instead, try to remember it is just a personal failure on their part -- it is, and the honest ones do admit to it --, prove them wrong with rock solid evidence, and move on.

    About my "stalkerish" response: perhaps some background is needed for others to understand what happened. I personally continuously look for interesting problems to solve, to hone my skills, and to learn new stuff. I still lurk here occasionally to do that, although I prefer not to participate anymore, for the simple reason that most of the regular posters here are unable to admit their own errors, simply ignoring any evidence otherwise, leaving other readers to decide for themselves -- and that irks me even more than dismissive maintainers. I lurk, because I haven't found a way to get over it. Computational materials physics being my field, Fortran and number crunching is familiar to me, and I found the topic very, very interesting. Any bugs that affect Binary128 floating-point implementation (in Fortran or C) is likely to bite me too in the future, so since I had some free time at my hands, I decided to jump in. The data in the first post was easily sufficient to locate the relevant GCC bugzilla entry, that's all. It's not like I'm deliberately hiding my own identity either; I just prefer to use a pseudonym.

    Which, by the way, is frowned upon in many open source projects. That is why I participate in open source projects using my real name, even if I would definitely prefer to use my pseudonym. I always recommend using your real name in open source projects, when entering a discussion in a mailing list or bugzilla; using a pseudonym is almost always enough reason for an outright dismissal. So, use your real name, and try to ignore those abrasive personality traits or persons, and provide as much reproducible evidence as you can, and it should be worth your while.

  12. #12
    misoturbutc Hodor's Avatar
    Join Date
    Nov 2013
    Posts
    1,791
    Is looking up a bug report stalking is it? Consider me guilty then. I found the thread in question and in all honesty I think that if you (Epy) had responded differently then the entire thread may have been less "abrasive". Just my opinion.

  13. #13
    Master Apprentice phantomotap's Avatar
    Join Date
    Jan 2008
    Posts
    5,108
    When I pointed out that I had tested on GNU/Linux plain GCC, I was met with hostility by this individual and he basically made every attempt to make me look dumb instead of being helpful. Basically made me feel like an .............. when I went out of my way to try and help the project by reporting the bug.
    O_o

    I love the community proper, but I hate a lot in the community. You offer a related solution that may be a better fit, and some people see it as an attack on a particular solution. You offer a possible alternative, and some people see it as a personal attack on their abilities. You offer a bug fix, and some people expend massive effort showing "PEBKAC" before considering the possibility of a bug.

    I was "harshed" by a few G++ developers for so long after one report--which eventually turned out to be a regression--that these days I use a burner email to discuss any bug just I can toss it rather than deal with the crap if the developers decide to see an attack.

    *shrug*

    But if it helps, your response wasn't as bad as I've seen. I once submitted a bug fix (I didn't just file a report. I offered a patch that fixed the issue without any regression shown by their own tests.) to a "Linux" module in C++. So what was my crime? Effectively, the C++ `integer_typedef label(get_a_value());' instead of the C `integer_typedef label = get_a_value();'. From the response, I may as well have personally raped the children of every developer.

    *sigh*

    I do also agree that "GNU" and "Linux" projects seem to be the most problematic, but you shouldn't let that discourage you from other projects. I've mostly found smaller teams much more receptive, also appreciative, of such feedback.

    Soma
    “Salem Was Wrong!” -- Pedant Necromancer
    “Four isn't random!” -- Gibbering Mouther

  14. #14
    Registered User
    Join Date
    Jun 2005
    Posts
    6,815
    Open source projects aren't the only source of developers with egos that run rampant when faced with a bug report.

    I've seen similar things with commercial developers, as well as with academics. On occasion, as contracted technical authority representing a paying customer (i.e. I have been delegated authority to prevent contracted payments) I've copped similar behaviour.


    The problem is rampant human ego, not project type, development methodology, or anything else.
    Right 98% of the time, and don't care about the other 3%.

    If I seem grumpy or unhelpful in reply to you, or tell you you need to demonstrate more effort before you can expect help, it is likely you deserve it. Suck it up, Buttercup, and read this, this, and this before posting again.

  15. #15
    Make Fortran great again
    Join Date
    Sep 2009
    Posts
    1,413
    Quote Originally Posted by Hodor View Post
    Is looking up a bug report stalking is it? Consider me guilty then. I found the thread in question and in all honesty I think that if you (Epy) had responded differently then the entire thread may have been less "abrasive". Just my opinion.
    I went for the throat because I'm so sick of people automatically assuming it's a mingw problem just for mentioning mingw. I'm still ........ed, I went out of my way to help the project by reporting the problem only to be immediately dismissed. If you're not going to read the bug report then don't bother.

Popular pages Recent additions subscribe to a feed

Similar Threads

  1. Open Source Beginner Projects
    By jingles in forum C# Programming
    Replies: 1
    Last Post: 04-12-2012, 10:00 AM
  2. Open Source projects in C
    By suryak in forum C Programming
    Replies: 5
    Last Post: 09-28-2011, 06:33 AM
  3. gigantic open source projects, seems unmanageable ?
    By rodrigorules in forum C Programming
    Replies: 3
    Last Post: 06-07-2010, 06:35 PM
  4. Two New C++ Open Source Projects
    By BigAngryDog in forum Projects and Job Recruitment
    Replies: 0
    Last Post: 05-01-2010, 07:01 PM
  5. Are these good ideas for open source projects?
    By darsunt in forum General Discussions
    Replies: 2
    Last Post: 03-26-2010, 05:04 PM