- With " warnings as errors ", you don't want it to stop the debuging. - 5 Updates
- May.25.2017 -- A plea for help - 4 Updates
- "Info" as a name - 2 Updates
- What is also the difference between animals and humans ? - 1 Update
| Bo Persson <bop@gmb.dk>: May 25 02:31AM +0200 On 2017-05-24 22:29, Robert Wessel wrote: > except under the "Visual C++" name. These have all, of course, had a > C mode. NT 3.51 DD development would likely have been done with MSVC > 1.5 or 2.0. At least at the time of NT you could get a Device Driver Kit with only the command line compiler and no Visual Studio. Same compiler though. Bo Persson |
| Robert Wessel <robertwessel2@yahoo.com>: May 24 07:55PM -0500 >At least at the time of NT you could get a Device Driver Kit with only >the command line compiler and no Visual Studio. >Same compiler though. IIRC, VS didn't exist as a product at the time 3.51 was released, and each of the languages came with (more or less) its own IDE. OTOH, I simply don't remember how the DDK was packaged back them I did install several, but if they came with the C compiler built in or added on to a Visual C++ installation, I have no memory. |
| scott@slp53.sl.home (Scott Lurndal): May 25 12:44PM >except under the "Visual C++" name. These have all, of course, had a >C mode. NT 3.51 DD development would likely have been done with MSVC >1.5 or 2.0. To be fair, I did all my editing using vi on an SVR3 box, and just copied the source over to the NT machine to compile and test them. The drivers were for WORM (optical, pre-CD) drives and a WORM jukebox. |
| David Brown <david.brown@hesbynett.no>: May 25 10:39PM +0200 On 24/05/17 17:05, Alf P. Steinbach wrote: >> unnecessary. > No, the list was compiled with input from a much larger than noe clc++ > community, at the time. (Scott, I read "noe" as "now", in the sense of "current". "noe" /is/ a Norwegian word, meaning "something", but that would not make sense in the context.) > Some sillywarnings were added later. > I have somewhere a ditto header for g++, but /that/ one is just my > personal opinion. In addition to gcc's -Wall and -Wextra, I explicitly turn /on/ several warnings that are the direct equivalent of some of the ones you are disabling here. Some of the others I am less sure about, and many do not have direct gcc equivalents. Note that this is primarily for my own code - third-party code often does not stick to the same strict standards as I do, and so I have to loosen my warnings for them. For example, I think it is a /good/ thing to check that all enum values are handled in a switch. That means that if I add a new value to an enum, and I forget to handle it in a switch, I get a warning. Each bug spotted by the compiler is a bug less to find in testing. I don't put functions in my code that are not used - I write for small embedded systems, and don't want to waste space. If I mark a function as "inline", and it is not inlined, I want to know about it. (But I don't care if a normal function /is/ inlined by the compiler.) If padding is added to a struct, I want to know about it. It is probably fair to say that disabling most of the warnings on your list would not be a problem to me. It is unlikely that they would be triggered by my code (after all, most of what I write is C rather than C++!). And I think if, for example, a copy constructor cannot be generated for a class, then that is because of the way you have written the class. It is only a problem if the copy constructor cannot be generated, and you need to use it - /that/ is when there needs to be a warning or error. |
| "Alf P. Steinbach" <alf.p.steinbach+usenet@gmail.com>: May 26 12:39AM +0200 On 25-May-17 10:39 PM, David Brown wrote: > not have direct gcc equivalents. Note that this is primarily for my own > code - third-party code often does not stick to the same strict > standards as I do, and so I have to loosen my warnings for them. I agree wholeheartedly with you that reasonable warnings should be turned on, in general. That was the purpose of the list: to allow one to use the VC compiler with the highest normal warning level, /W4, without getting bogged down in an avalanche of system header warnings. Today, if I were addressing this area I would make a header that /turns on/ some warnings that the below referenced Microsoft list says are turned off in VC 2017, in particular the signed/unsigned mismatch warning C4389. > are handled in a switch. That means that if I add a new value to an > enum, and I forget to handle it in a switch, I get a warning. Each bug > spotted by the compiler is a bug less to find in testing. This was, in 2010, a Visual C++ warning that generated false positives for code like enum class Enum{ a, b, c, d, e, f }; auto foo() -> Enum; auto main() -> int { switch( foo() ) { case Enum::a: return 10; case Enum::b: return 20; default: break; } return 30; } When one uses `default:` in a `switch` then that means that one is handling every other value. But Visual C++ didn't see it that way: the VC designers thought that it was a good idea to warn about the use of `default:`, for what if it had been placed there by, say, a copy/paste error, or sumthin'?. Hence in 2010 it was a negative value sillywarning. With Visual C++ 2017 that warning, C4061, has been /turned off/ by default. Presumably for the same reason as my sillywarnings-list. See the list at <https://docs.microsoft.com/en-us/cpp/preprocessor/compiler-warnings-that-are-off-by-default>; as it happens this warning is on top. I think this shows that for this particular concrete example you're arguing from ignorance, due to simply not having much experience with this compiler and its idiosyncrasies. > If I mark a function as "inline", and it is not inlined, I want to know > about it. (But I don't care if a normal function /is/ inlined by the > compiler.) `inline`'s effect on inlining of calls is just a hint. Its main purpose, and its only guaranteed effect, is to ensure that a function can be defined in a header; technically, that it can be defined in the same way, with extern linkage, in multiple translation units. When the keyword is used for that guaranteed effect, or when a member function is defined in the class definition (which implicitly gives it `inline`), you really don't want the compiler spewing out diagnostics about ignoring the additional vague hint semantics. So here you're evidently doing something wrong. Where you want an all-out effort to inline a call (I think it's better to let the compiler decide, but let's say profiling says that the compiler didn't inline where it should, and that it's critical), use the appropriate compiler specific magic invocation. Standard C++ doesn't have that feature; relying on `inline` to do /that/ job is not portable. But the compiler-specific can be wrapped in a macro for portability. This warning, C4710, is also off by default in Visual C++ 2017. > If padding is added to a struct, I want to know about it. No, really, you don't. Padding is added to most structs. Only if you have designated the struct as packed, is such a warning useful. And also this warning, C4820, is one of those that now are off by default. I.e. for the three warnings so far, Microsoft have, wisely IMHO, agreed with the sillywarnings list. :-) > C++!). And I think if, for example, a copy constructor cannot be > generated for a class, then that is because of the way you have written > the class. No, that could be e.g. `boost::noncopyable`. > It is only a problem if the copy constructor cannot be > generated, and you need to use it - /that/ is when there needs to be a > warning or error. Yes, I agree, but Visual C++ has had a habit of emitting these diagnostics whenever, as it looks from a purely local context ignoring everything else, they /might/ apply. Cheers & hth., - Alf |
| "Rick C. Hodgin" <rick.c.hodgin@gmail.com>: May 25 04:52AM -0700 This world is run by evil madmen being guided by an insane evil spirit named Satan. The purpose of all of Satan's actions are to do the opposite of that which God established, and to teach contrary things to that which God established. I would like to request help in changing that. The Lord Jesus Christ came to this Earth to restore that which was lost because of sin. He did that at the cross, having fulfilled all He was tasked to do completely, and faithfully, even though it was a most bitter cup He was forced to drink from. We now in this world have a choice. We can continue on in the way and ways of the world, or we can recognize them for what they are (evil), and choose to go another way. I am asking each of you to consider the direction you want this world to go in, and make a conscious choice to put your talents and skills where your mouth is, and step up in this world and serve the Lord Jesus Christ directly with your life's focus. Together, we can achieve that which none of us can achieve on our own. I offer my talents and skills and knowledge and resources unto the Lord, and I'm asking you to join with me in moving forward with projects that seek to honor Him in this world. Consciously. Purposefully. By a concerted, focused effort desiring to turn our backs on the evil ways of this world, and embrace the holy, right, and true ways taught by the Lord Jesus Christ. Everything He has taught us and done for us is right, pure, holy, and that very and exact thing we all need in our lives. I choose to embrace fully His presence in my life, and to bring Him out front ahead of the things I do, so that I am not doing them for myself, for money, for some kind of fame or glory, but I am instead doing it because I recognize He first made me, and gave me my skills, and gave me every opportunity I've had to move in this world. And I want to give back to Him that which He first gave me, as an offering of my free will because of who He is. I invite each of you to step up and come and join this project I am working on. It has been a monumental undertaking, and I have been diminished in my ability to continue working on it over time. I am not sure the cause of that diminishment, but I know that together we can achieve that which He first placed upon my heart: A full hardware and software stack, with our own protocols and specifications that are unencumbered by existing patents and IP constraints: (1) CPU (2) Mainboard components (3) Memory (4) Networking, video, sound (5) Operating system (6) Drivers (7) Tools (editors, compilers, debuggers, script engines) (8) Applications (9) Physical manufacturing of all of these things We have the skills. We have the talent. We can tear down and rebuild that which the enemy constructed on false foundations. We can make ours fully secure upon eternal foundations. I am asking you to come forth with your life's offering and serve the Lord. I'm asking you to come to His Son and ask forgiveness of your sin and then begin your life anew in service to Him. The enemy won't like this, and he will come and attack from all sides. But God is far greater, and if you stay the course, God Himself will ensure your success ... because of who He is, and because of who you are in Him. Thank you for considering this proposal. Feel free to email me if you would like. I look forward to hearing from you. Thank you, Rick C. Hodgin |
| "Rick C. Hodgin" <rick.c.hodgin@gmail.com>: May 25 06:36AM -0700 On Thursday, May 25, 2017 at 7:52:27 AM UTC-4, Rick C. Hodgin wrote: > are in Him. > Thank you for considering this proposal. Feel free to email me if > you would like. I look forward to hearing from you. Someone in another group posted this: ----- In comp.arch.fpga: On Thursday, May 25, 2017 at 9:05:23 AM UTC-4, Evgeny Filatov wrote: > > (8) Applications > > (9) Physical manufacturing of all of these things > So... what has been your progress so far? They're all works in progress at various stages of completion. CPU design: http://www.libsf.org:8990/projects/LIB/repos/libsf/browse/arxoda/core FPGA design: http://www.libsf.org:8990/projects/LIB/repos/libsf/browse/arlina Video hardware: http://www.libsf.org:8990/projects/LIB/repos/libsf/browse/arxoda/core/video/video_controller.png OS kernel: http://www.libsf.org:8990/projects/LIB/repos/libsf/browse/exodus/source http://www.visual-freepro.org/videos/2014_02_13__exodus_debi_debugger.ogv C-like software language: https://groups.google.com/forum/#!forum/caliveprogramminglanguage https://groups.google.com/forum/#!forum/rapid_development_compiler http://www.libsf.org:8990/projects/LIB/repos/libsf/browse/exodus/tools And I have other apps I've worked on, and hardware: Font system: http://www.libsf.org:8990/projects/LIB/repos/libsf/browse/utils/dsf Database system: http://www.libsf.org:8990/projects/LIB/repos/libsf/browse/source/vjr Automotive fuel injection system, with in-progress simulation: http://www.libsf.org:8990/projects/LIB/repos/libsf/browse/source/superjet http://www.libsf.org/images/superjet/superjet_feb_18_2017.avi IBM Model-F buckling spring style keyboard manufacturing: http://www.libsf.org:8990/projects/LIB/repos/libsf/browse/king http://www.libsf.org:8990/projects/LIB/repos/libsf/browse/king/keyboard_designs.png And I have other ideas and projects mapped out, including a non-linear on-the-fly video / audio editor called Whitebox, the purpose of which is to apply user edits to raw content decoded from a DVD, so you pop in a DVD which is an R-rated movie due to profanity and/or violence, but one which otherwise has a good message and good scenes, and by applying on-the-fly edits, you wind up with a G-rated version suitable for the family. And I have ideas for how to create a Blender-like app, a GIMP-like app, as well as a full office suite, etc. Thank you, Rick C. Hodgin |
| bitrex <bitrex@de.lete.earthlink.net>: May 25 04:04PM -0400 On 05/25/2017 09:36 AM, Rick C. Hodgin wrote: >> named Satan. The purpose of all of Satan's actions are to do the >> opposite of that which God established, and to teach contrary things >> to that which God established. Sounds about right > app, a GIMP-like app, as well as a full office suite, etc. > Thank you, > Rick C. Hodgin How are the cut points determined? The reason most movies that are rated "R" get that rating is because there was so much violence they couldn't edit it down to "PG-13" and have it still make sense. That or someone showed a boob once for 3 seconds. I can't think of many films worth watching by anyone, family or otherwise, in the past 20 years that had a "good message" that also got an "R" rating. Saving Private Ryan, maybe. I swore off Hollywood dribble 99% many years ago, feels great without it |
| "Rick C. Hodgin" <rick.c.hodgin@gmail.com>: May 25 01:50PM -0700 On Thursday, May 25, 2017 at 4:04:36 PM UTC-4, bitrex wrote: > > for the family. And I have ideas for how to create a Blender-like > > app, a GIMP-like app, as well as a full office suite, etc. > How are the cut points determined? The databases that are applied to the live Whitebox edits are provided by the community. They can be downloaded, options selected, new edits added, unwanted edits removed, etc., and they go into a database where people can search and make comments and notes. All of this is spec'd out, but none of it exists yet. My whole goal was to allow modified editing of purchased content, so the movie studios have no reason to complain because they're still getting money for their movie, and people are able to see/hear only what they want to. > I can't think of many films worth watching by anyone, family or > otherwise, in the past 20 years that had a "good message" that > also got an "R" rating. Saving Private Ryan, maybe. Most films have enough content in them I don't want to watch them either (G on up). Some entire themes need to be removed. The same is true for TV shows, and Whitebox would work on TV shows. I like many stories and the writing on The West Wing writing, Numb3rs, but they have too many sub-themes which are inappropriate. I would Whitebox those out entirely. Even shows like Star Trek Enterprise, edit out every decon chamber scene, every sexual reference, every neural pressure therapy scene between T'Pol and Trip, and so on. And the list goes on. In most cases, remove up to 20% of the content, and it's then a G-rated show with all sub-themes removed, suitable for the family. This is on a movie that is already family-targeted, of course. It wouldn't work with The Terminator or other such movie. > I swore off Hollywood dribble 99% many years ago, feels great > without it Me too. But there are some movies I would like to see again, but can't watch because of language or various few scenes. I have manually edited some movies down to have just the G-rated content. Two in particular are: Real Genius 1985: http://www.imdb.com/title/tt0089886/ By The Sword 1991: http://www.imdb.com/title/tt0101524/ And I'm working on: Ice Station Zebra 1968: http://www.imdb.com/title/tt0063121/ The American President: http://www.imdb.com/title/tt0112346/ I also plan on working The West Wing seasons 1-4, Numb3rs, Sports Night, and possibly Northern Exposure. But, those are really far- off long-term plans that I will get to someday (if I do). Thank you, Rick C. Hodgin |
| Tim Rentsch <txr@alumni.caltech.edu>: May 25 06:51AM -0700 > Your version is less obvious to me, what with the recursion and > general indirection, even though I've read a few Lisp and functional > programming books [...]. Delegating constructors fit nicely into a functional/recursive approach, but they have an important limitation in that they must be unconditional. That means we must have a helper function in some cases to calculate values for the member variables. Here the code is taking advantage of the property that once the number of leading ones is known, it's easy to mask off the leading ones from the original parameter value, using a simple expression. That leaves the recursive function high_ones(). I think it's pretty easy to read functions like this once you get the hang of it. Case in point: if we see this return bits & mask ? high_ones( n+1, bits, mask>>1 ) : n; we might think of it as a loop while ( bits & mask ) { n += 1, mask >>= 1; } return n; which may help bridge the gap until the recursive writing becomes more natural. One advantage of adopting a functional/recursive approach is that it provides a way of "looping" in functions but which still can be 'constexpr'. > functional programming like in Lisp), it's the precedence of the C++ > `^`, `<<` and infix `-` operators. I'd had to look that up to be > sure! Here's an easy way to remember (for operators that are also in C, which holds in this case). Binary operators all group in ways that are natural and sensible given what they do, except for the bitwise operators. Bitwise operators form a "sandwich" around the relational/equality operators: shift operators have higher precedence, and bitwise-logic operators have lower precedence. The two kinds of shift have the same precedence; the three kinds of bitwise-logic operations each have their own level: & and | parallel && and ||, and ^ is inbetween. The logic operations having lower precedence than relational/equality operators is a historical artefact of early C not having && and ||, so it made sense (then, not now) for & and | to be "lower" than ==, etc, for boolean combination of conditions. Here is a variation on the previous class definition - the helper function calculates both member values at once, packing them into a single 'unsigned long long' value for subsequent extraction. Notice the private constructor needs to be 'explicit' in this case, to avoid an overloading ambiguity. typedef unsigned char Byte; struct EncodedByte { int const n_leading_ones; Byte const low_bits; constexpr EncodedByte( char c ) : EncodedByte( two_parts( 0, c & 255u, 0x80 ) ) {} private: typedef unsigned Bits; typedef unsigned long long ULL; explicit constexpr EncodedByte( ULL upper_lower ) : n_leading_ones( upper_lower >> 8 ) , low_bits( upper_lower & 255 ) {} static constexpr ULL two_parts( ULL n, Bits b, Bits m ){ return b & m ? two_parts( n+1, b^m, m>>1 ) : n<<8 | b; } }; |
| "Alf P. Steinbach" <alf.p.steinbach+usenet@gmail.com>: May 25 04:30PM +0200 On 25-May-17 3:51 PM, Tim Rentsch wrote: > One advantage of adopting a functional/recursive approach is that > it provides a way of "looping" in functions but which still can > be 'constexpr'. Oh, this was relaxed in C++14. E.g. the following code compiles fine with g++ (but Visual C++ 2017 lacks support): constexpr auto sum_to( int const n ) -> int { int sum = 0; for( int i = 1; i <= n; ++i ) { sum += i; } return sum; } #include <stdlib.h> auto main() -> int { constexpr int n = 42; char const a[sum_to( n )]{}; return (sizeof(a) == n*(n + 1)/2? EXIT_SUCCESS : EXIT_FAILURE); } Thanks for your comments, third approach for the problem, and further help about how to remember operator precedence (snipped here). :) Cheers!, - Alf |
| Bonita Montero <Bonita.Montero@gmail.com>: May 25 10:11AM +0200 According to your posing-length you must be severely manic. Go to a doctor. |
| You received this digest because you're subscribed to updates for this group. You can change your settings on the group membership page. To unsubscribe from this group and stop receiving emails from it send an email to comp.lang.c+++unsubscribe@googlegroups.com. |
No comments:
Post a Comment