On Wed, 03 Jun 2015 11:35:41 +0000, weaselcat wrote:

> On Wednesday, 3 June 2015 at 11:25:50 UTC, Ola Fosheim Grøstad wrote:
>> On Wednesday, 3 June 2015 at 11:06:39 UTC, weaselcat wrote:
>>> it can get even better if you properly modularize your projects instead of having 1-2 files that build slow, which causes a lot of waiting.
>>
>> Yes, sure. You can probably get the same build speeds as with C if you organize your code in a particular way or shy away from certain patterns.
>>
>> What I was talking about was the language, meaning that you don't write your code to give a boost in compilation speed. Clearly possible, but if you use third party frameworks… then you're out of luck.
>>
>> An analogy: SQL isn't particularly expressive, but the limitations of the language makes it possible to execute it bottom-up. NOSQL engines are even less expressive, but can be even more easily distributed.
> 
> ah yes, those famous fast C build times.
> Excuse me while I go take half an hour to build GDB.

yet gcc (the C compiler part) is significantly faster than gdc on my box (both gcc and gdc are built from sources, tuned to my arch). i mean separate compilation, of course. i believe it's due to phobos.

On Wed, 03 Jun 2015 12:20:28 +0000, Paulo  Pinto wrote:

>> Now really? C was designed at a time where you couldn't even hold the source file in memory, so there is not even a need for an explicit AST.
>>
>> C can essentially be "streamed" in separate passes: cpp->cc->asm->linking
>>
>> If compiling C is slow, it is just the compiler or the build system, not the language.
> 
> Yes really, specially when comparing with Turbo Pascal, Delphi, Modula-2, Oberon and a few other languages not tied to UNIX linker model.

yes, i remember lightning fast compile times with turbo pascal. yet the code it produced was really awful: it was even unable to fold constants sometimes!

ketmar <ketmar@ketmar.no-ip.org> writes:

> yes, i remember lightning fast compile times with turbo pascal. yet the code it produced was really awful: it was even unable to fold constants sometimes!

I remember it being in a single DOS .COM (was it TURBO.COM?) only about 40k which included the editor, compiler, and libraries.  It was the coolest thing for PC's at the time.  I might even have a floppy with it on it :-)

On Thursday, 4 June 2015 at 03:04:31 UTC, ketmar wrote:
> On Wed, 03 Jun 2015 12:20:28 +0000, Paulo  Pinto wrote:
>
>>> Now really? C was designed at a time where you couldn't even hold the
>>> source file in memory, so there is not even a need for an explicit AST.
>>>
>>> C can essentially be "streamed" in separate passes:
>>> cpp->cc->asm->linking
>>>
>>> If compiling C is slow, it is just the compiler or the build system,
>>> not the language.
>> 
>> Yes really, specially when comparing with Turbo Pascal, Delphi,
>> Modula-2, Oberon and a few other languages not tied to UNIX linker
>> model.
>
> yes, i remember lightning fast compile times with turbo pascal. yet the
> code it produced was really awful: it was even unable to fold constants
> sometimes!

No different from other MS-DOS C compilers.

Hence why such languages were the Pythons and Rubys of the day
and anyone that cared about performance was using straight
Assembly, in MS-DOS and other home systems, that is.

Michael Abrash books The Zen of Assembly Language and  Zen of
Code Optimization were published in 1990 and 1994 respectively.

--
Paulo

On 2015-06-03 12:37, "Ola Fosheim =?UTF-8?B?R3LDuHN0YWQi?= <ola.fosheim.grostad+dlang@gmail.com>" wrote:

> Now really? C was designed at a time where you couldn't even hold the
> source file in memory, so there is not even a need for an explicit AST.
>
> C can essentially be "streamed" in separate passes: cpp->cc->asm->linking
>
> If compiling C is slow, it is just the compiler or the build system, not
> the language.

Doesn't a C compiler need to reparse headers in C? Unlike D were they can be cached.

-- 
/Jacob Carlborg

On 29/05/2015 19:35, H. S. Teoh via Digitalmars-d wrote:
> [...]
>> >This isn't the first time I've seen this, in basically every benchmark
>> >featuring D I have to submit a patch/make a comment that dmd shouldn't
>> >be used. Make no mistake, this is damaging to D's reputation - how
>> >well does D's "native efficiency" go over when people are saying it's
>> >slower than Scala and F#(mono)?
>> >
>> >LDC and GDC need promoted more.
> [...]
>
> This will probably offend some people, but I think LDC/GDC should be the
> default download on dlang.org, and dmd should be provided as an
> alternative for those who want the latest language version and don't
> mind the speed compromise.

It should be more than just LDC/GDC being the default download on dlang.org, the DM backend and related toolchain should be phased out altogether in favor of LLVM.

Walter might have written great compiler tools in the 90s or so, but in today's internet and FOSS online-collaborative era, how can the Digital Mars toolchain hope to compete with toolchains having teams of multiple full-time developers working on it? (plus a plethora of occasional volunteer contributors). The difference in manpower and resources is astonishing! And it's only gonna get bigger since LLVM is having more and more people and companies supporting it. By this rate, it may well one day make even GCC old and obsolete, left to be used by FSF zealots only.

-- 
Bruno Medeiros
https://twitter.com/brunodomedeiros

On 5 Jun 2015 20:55, "Bruno Medeiros via Digitalmars-d" < digitalmars-d@puremagic.com> wrote:
>
> On 29/05/2015 19:35, H. S. Teoh via Digitalmars-d wrote:
>>
>> [...]
>>
>>> >This isn't the first time I've seen this, in basically every benchmark featuring D I have to submit a patch/make a comment that dmd shouldn't be used. Make no mistake, this is damaging to D's reputation - how well does D's "native efficiency" go over when people are saying it's slower than Scala and F#(mono)?
>>> >
>>> >LDC and GDC need promoted more.
>>
>> [...]
>>
>>
>> This will probably offend some people, but I think LDC/GDC should be the default download on dlang.org, and dmd should be provided as an alternative for those who want the latest language version and don't mind the speed compromise.
>
>
> It should be more than just LDC/GDC being the default download on
dlang.org, the DM backend and related toolchain should be phased out altogether in favor of LLVM.
>
> Walter might have written great compiler tools in the 90s or so, but in
today's internet and FOSS online-collaborative era, how can the Digital Mars toolchain hope to compete with toolchains having teams of multiple full-time developers working on it? (plus a plethora of occasional volunteer contributors). The difference in manpower and resources is astonishing! And it's only gonna get bigger since LLVM is having more and more people and companies supporting it. By this rate, it may well one day make even GCC old and obsolete, left to be used by FSF zealots only.
>

At the risk of speaking with lack of foresight, are you on the gcc mailing list too? If not, get on it. Otherwise you will enter this kind of polarised view of X will dominate all.

On Friday, 5 June 2015 at 22:07:48 UTC, Iain Buclaw wrote:
> On 5 Jun 2015 20:55, "Bruno Medeiros via Digitalmars-d" <
> digitalmars-d@puremagic.com> wrote:
>>
>> On 29/05/2015 19:35, H. S. Teoh via Digitalmars-d wrote:
>>>
>>> [...]
>>>
>>>> >This isn't the first time I've seen this, in basically every benchmark
>>>> >featuring D I have to submit a patch/make a comment that dmd shouldn't
>>>> >be used. Make no mistake, this is damaging to D's reputation - how
>>>> >well does D's "native efficiency" go over when people are saying it's
>>>> >slower than Scala and F#(mono)?
>>>> >
>>>> >LDC and GDC need promoted more.
>>>
>>> [...]
>>>
>>>
>>> This will probably offend some people, but I think LDC/GDC should be the
>>> default download on dlang.org, and dmd should be provided as an
>>> alternative for those who want the latest language version and don't
>>> mind the speed compromise.
>>
>>
>> It should be more than just LDC/GDC being the default download on
> dlang.org, the DM backend and related toolchain should be phased out
> altogether in favor of LLVM.
>>
>> Walter might have written great compiler tools in the 90s or so, but in
> today's internet and FOSS online-collaborative era, how can the Digital
> Mars toolchain hope to compete with toolchains having teams of multiple
> full-time developers working on it? (plus a plethora of occasional
> volunteer contributors). The difference in manpower and resources is
> astonishing! And it's only gonna get bigger since LLVM is having more and
> more people and companies supporting it. By this rate, it may well one day
> make even GCC old and obsolete, left to be used by FSF zealots only.
>>
>
> At the risk of speaking with lack of foresight, are you on the gcc mailing
> list too? If not, get on it. Otherwise you will enter this kind of
> polarised view of X will dominate all.

 Slightly off topic, but I recently started digging into GDC( on your personal fork, anyways.) I find the code pleasantly easy to navigate and understand. I don't think I've given gdc its due credit in this thead.

Bye,

On 6 June 2015 at 01:20, weaselcat via Digitalmars-d < digitalmars-d@puremagic.com> wrote:

> On Friday, 5 June 2015 at 22:07:48 UTC, Iain Buclaw wrote:
>
>> On 5 Jun 2015 20:55, "Bruno Medeiros via Digitalmars-d" < digitalmars-d@puremagic.com> wrote:
>>
>>>
>>> On 29/05/2015 19:35, H. S. Teoh via Digitalmars-d wrote:
>>>
>>>>
>>>> [...]
>>>>
>>>>  >This isn't the first time I've seen this, in basically >every
>>>>> benchmark
>>>>> >featuring D I have to submit a patch/make a comment that >dmd
>>>>> shouldn't
>>>>> >be used. Make no mistake, this is damaging to D's >reputation - how well does D's "native efficiency" go over when people are >saying it's slower than Scala and F#(mono)?
>>>>> >
>>>>> >LDC and GDC need promoted more.
>>>>>
>>>>
>>>> [...]
>>>>
>>>>
>>>> This will probably offend some people, but I think LDC/GDC should be the default download on dlang.org, and dmd should be provided as an alternative for those who want the latest language version and don't mind the speed compromise.
>>>>
>>>
>>>
>>> It should be more than just LDC/GDC being the default download on
>>>
>> dlang.org, the DM backend and related toolchain should be phased out altogether in favor of LLVM.
>>
>>>
>>> Walter might have written great compiler tools in the 90s or so, but in
>>>
>> today's internet and FOSS online-collaborative era, how can the Digital Mars toolchain hope to compete with toolchains having teams of multiple full-time developers working on it? (plus a plethora of occasional volunteer contributors). The difference in manpower and resources is astonishing! And it's only gonna get bigger since LLVM is having more and more people and companies supporting it. By this rate, it may well one day make even GCC old and obsolete, left to be used by FSF zealots only.
>>
>>>
>>>
>> At the risk of speaking with lack of foresight, are you on the gcc mailing list too? If not, get on it. Otherwise you will enter this kind of polarised view of X will dominate all.
>>
>
>  Slightly off topic, but I recently started digging into GDC( on your
> personal fork, anyways.) I find the code pleasantly easy to navigate and
> understand. I don't think I've given gdc its due credit in this thead.
>
> Bye,
>

If you've been following the 2.067 re-work, that is really the way things are going right now.  More encapsulation, less of the flat hierachal structure is the key securing future interest.  My hope is that GDC will fall into the "what a good frontend shoudl do" category after I'm done.

On 05/06/2015 23:07, Iain Buclaw via Digitalmars-d wrote:
>
> On 5 Jun 2015 20:55, "Bruno Medeiros via Digitalmars-d"
> <digitalmars-d@puremagic.com <mailto:digitalmars-d@puremagic.com>> wrote:
>  >
>  > On 29/05/2015 19:35, H. S. Teoh via Digitalmars-d wrote:
>  >>
>  >> [...]
>  >>
>  >>> >This isn't the first time I've seen this, in basically every benchmark
>  >>> >featuring D I have to submit a patch/make a comment that dmd shouldn't
>  >>> >be used. Make no mistake, this is damaging to D's reputation - how
>  >>> >well does D's "native efficiency" go over when people are saying it's
>  >>> >slower than Scala and F#(mono)?
>  >>> >
>  >>> >LDC and GDC need promoted more.
>  >>
>  >> [...]
>  >>
>  >>
>  >> This will probably offend some people, but I think LDC/GDC should be the
>  >> default download on dlang.org <http://dlang.org>, and dmd should be
> provided as an
>  >> alternative for those who want the latest language version and don't
>  >> mind the speed compromise.
>  >
>  >
>  > It should be more than just LDC/GDC being the default download on
> dlang.org <http://dlang.org>, the DM backend and related toolchain
> should be phased out altogether in favor of LLVM.
>  >
>  > Walter might have written great compiler tools in the 90s or so, but
> in today's internet and FOSS online-collaborative era, how can the
> Digital Mars toolchain hope to compete with toolchains having teams of
> multiple full-time developers working on it? (plus a plethora of
> occasional volunteer contributors). The difference in manpower and
> resources is astonishing! And it's only gonna get bigger since LLVM is
> having more and more people and companies supporting it. By this rate,
> it may well one day make even GCC old and obsolete, left to be used by
> FSF zealots only.
>
> At the risk of speaking with lack of foresight, are you on the gcc
> mailing list too? If not, get on it. Otherwise you will enter this kind
> of polarised view of X will dominate all.
>

I'm not on any LLVM mailing list or forum either. It's too much volume for what I need to care/know about. I'm only on the LLVM Weekly newsletter. If GCC has a similar newsletter I might sign up to that.


-- 
Bruno Medeiros
https://twitter.com/brunodomedeiros