D - DMD 0.78 release
- Walter (2/2) Jan 14 2004 Mainly bug fixes.
- Matthew (3/5) Jan 14 2004 Hmm. I reckon it could do with a nice memory-mapped file IO module. ;)
- Sean L. Palmer (8/14) Jan 20 2004 Sure!
- Matthew (13/28) Jan 21 2004 Pleasure. Good to hear appreciation any time. :)
- ssuukk (3/7) Jan 14 2004 Since you're going (as I see) into heavy de-bugging, do you want any
- J Anderson (8/15) Jan 14 2004 That's one of the reasons for this group. That way other users can look
- ssuukk (4/12) Jan 14 2004 Well my bug with static struct pseudo-constructors was posted here. But
- J Anderson (7/19) Jan 14 2004 This is a binary newsgroup so if it's < 200kb (this is just a wild guess...
- Walter (9/13) Jan 14 2004 Pseudo-code doesn't work for me. Realistically, I'm not going to try and
- ssuukk (8/20) Jan 15 2004 Heh... In this case it is possible to reduce this problem to two files
- Walter (4/7) Jan 15 2004 The language follows a pretty uniform symbol lookup rule, it does not ig...
- Sean L. Palmer (10/23) Jan 20 2004 Yeah, and aren't the ones that fall into that other 2% a royal bitch!
- Walter (14/20) Jan 21 2004 Yup. Fortunately, they are rare.
Mainly bug fixes. http://www.digitalmars.com/d/changelog.html
Jan 14 2004
Hmm. I reckon it could do with a nice memory-mapped file IO module. ;) "Walter" <walter digitalmars.com> wrote in message news:bu31vg$6bm$1 digitaldaemon.com...Mainly bug fixes. http://www.digitalmars.com/d/changelog.html
Jan 14 2004
Sure! Hey, Matthew, thanks alot for making that recls library. I often want something like that but end up having to write it myself, in hacked, limited form, every time. for (all files mathing pattern in subdirectory, recursive) do something(); Sean "Matthew" <matthew.hat stlsoft.dot.org> wrote in message news:bu33kk$8pf$1 digitaldaemon.com...Hmm. I reckon it could do with a nice memory-mapped file IO module. ;) "Walter" <walter digitalmars.com> wrote in message news:bu31vg$6bm$1 digitaldaemon.com...Mainly bug fixes. http://www.digitalmars.com/d/changelog.html
Jan 20 2004
"Matthew" <matthew.hat stlsoft.dot.org> Pleasure. Good to hear appreciation any time. :) any .NET), Java and D at the moment. The next column - May - will cover COM, and the July one will do Python. Then it's Perl and Ruby, and who knows what after that ... In the future I plan to expand it to handle recursive FTP handling, but that's on the back of several higher-priorities. Matthew P.S. Any PySequence gurus out there in D land? I may be about to step out of my comfort zone ... "Sean L. Palmer" <palmer.sean verizon.net> wrote in message news:bul8so$2j4i$1 digitaldaemon.com...Sure! Hey, Matthew, thanks alot for making that recls library. I often want something like that but end up having to write it myself, in hacked,limitedform, every time. for (all files mathing pattern in subdirectory, recursive) do something(); Sean "Matthew" <matthew.hat stlsoft.dot.org> wrote in message news:bu33kk$8pf$1 digitaldaemon.com...Hmm. I reckon it could do with a nice memory-mapped file IO module. ;) "Walter" <walter digitalmars.com> wrote in message news:bu31vg$6bm$1 digitaldaemon.com...Mainly bug fixes. http://www.digitalmars.com/d/changelog.html
Jan 21 2004
Walter wrote:Mainly bug fixes. http://www.digitalmars.com/d/changelog.htmlSince you're going (as I see) into heavy de-bugging, do you want any pieces of code that cause problems? If yes, where should I send them?
Jan 14 2004
ssuukk wrote:Walter wrote:That's one of the reasons for this group. That way other users can look at your bug report and tell you if it is actually a bug. Also it helps reduce double reporting. I personally use this format: [Bug] The bug. because it makes them easier to find. -AndersonMainly bug fixes. http://www.digitalmars.com/d/changelog.htmlSince you're going (as I see) into heavy de-bugging, do you want any pieces of code that cause problems? If yes, where should I send them?
Jan 14 2004
That's one of the reasons for this group. That way other users can look at your bug report and tell you if it is actually a bug. Also it helps reduce double reporting. I personally use this format: [Bug] The bug. because it makes them easier to find.Well my bug with static struct pseudo-constructors was posted here. But I don't think I should post whole code (which consists of several file) that causes this bug to the newsgroup :-) And I guess Walter needs this code to replicate this problem...
Jan 14 2004
ssuukk wrote:This is a binary newsgroup so if it's < 200kb (this is just a wild guess on what is appropriate, this newsgroup can handle much larger files ~ 700kb) it should be ok to send it as an attachment. Otherwise send it to Walters email. Walter prefers it if you can reduce the problem down to the critical code. Anyway, the more bug reports the better. We've gotta get the compiler right.That's one of the reasons for this group. That way other users can look at your bug report and tell you if it is actually a bug. Also it helps reduce double reporting. I personally use this format: [Bug] The bug. because it makes them easier to find.Well my bug with static struct pseudo-constructors was posted here. But I don't think I should post whole code (which consists of several file) that causes this bug to the newsgroup :-) And I guess Walter needs this code to replicate this problem...
Jan 14 2004
"ssuukk" <ssuukk .go2.pl> wrote in message news:bu3op2$1aqh$1 digitaldaemon.com...Well my bug with static struct pseudo-constructors was posted here. But I don't think I should post whole code (which consists of several file) that causes this bug to the newsgroup :-) And I guess Walter needs this code to replicate this problem...Pseudo-code doesn't work for me. Realistically, I'm not going to try and recreate a large amount of code to replicate the problem. I need reproducible examples, so email them if they're large. Also, the smaller the source code can be whittled down to to show the problem, frankly, the more likely it is it will get fixed. In my experience, about 98% of problems can be cut down to 10 lines of code or less. Thanks!
Jan 14 2004
Heh... In this case it is possible to reduce this problem to two files (not less!) because it occurs only, when using modules! Single file compiles without problems, only when I include another module it reports error thinking that "myvector.make(xxx)" is module name instead of static function call! I SUSPECT that this is because the file name in which I define "myvector" struct is also "myvector.d", but still - compier should know when "myvector.make" is class path and when function call, or am I wrong?Well my bug with static struct pseudo-constructors was posted here. But I don't think I should post whole code (which consists of several file) that causes this bug to the newsgroup :-) And I guess Walter needs this code to replicate this problem...Pseudo-code doesn't work for me. Realistically, I'm not going to try and recreate a large amount of code to replicate the problem. I need reproducible examples, so email them if they're large. Also, the smaller the source code can be whittled down to to show the problem, frankly, the more likely it is it will get fixed. In my experience, about 98% of problems can be cut down to 10 lines of code or less.
Jan 15 2004
"ssuukk" <ssuukk .go2.pl> wrote in message news:bu5hd9$14ts$1 digitaldaemon.com...I SUSPECT that this is because the file name in which I define "myvector" struct is also "myvector.d", but still - compier should know when "myvector.make" is class path and when function call, or am I wrong?The language follows a pretty uniform symbol lookup rule, it does not ignore symbols even if they are obviously wrong.
Jan 15 2004
Yeah, and aren't the ones that fall into that other 2% a royal bitch! Occasionally a problem will be such that making a small reprodicible sample would be 100x more work than you hitting the code in the debugger, seeing the obvious and writing a quick fix. Sometimes you may never be able to send a reproducible sample (machine-specific wierdness) Sean "Walter" <walter digitalmars.com> wrote in message news:bu44bb$1u2r$1 digitaldaemon.com..."ssuukk" <ssuukk .go2.pl> wrote in message news:bu3op2$1aqh$1 digitaldaemon.com...theWell my bug with static struct pseudo-constructors was posted here. But I don't think I should post whole code (which consists of several file) that causes this bug to the newsgroup :-) And I guess Walter needs this code to replicate this problem...Pseudo-code doesn't work for me. Realistically, I'm not going to try and recreate a large amount of code to replicate the problem. I need reproducible examples, so email them if they're large. Also, the smallersource code can be whittled down to to show the problem, frankly, the more likely it is it will get fixed. In my experience, about 98% of problemscanbe cut down to 10 lines of code or less. Thanks!
Jan 20 2004
"Sean L. Palmer" <palmer.sean verizon.net> wrote in message news:bul93s$2jec$1 digitaldaemon.com...Yeah, and aren't the ones that fall into that other 2% a royal bitch!Yup. Fortunately, they are rare.Occasionally a problem will be such that making a small reprodiciblesamplewould be 100x more work than you hitting the code in the debugger, seeing the obvious and writing a quick fix.I still need it reduced down. One reason is that after I fix it, it gets adapted into the D test suite so it never rears its nasty head again.Sometimes you may never be able to send a reproducible sample (machine-specific wierdness)That can happen with some kinds of products, but rarely with batch-oriented programs like compilers. From a pragmatic point of view, I'm much more likely to spend time working to fix an obvious problem reproducible in a few lines than a vague problem that comes with 300k of unfamiliar source across dozens of files and some complicated build process that will take a lot of time to figure out. My plate is pretty full <g>, and I am forced to triage things into what's the best return on the time invested.
Jan 21 2004