Re: x86 is stone age, how about moving along ?

[Andy Levy <an...@gmail.com>]

On Sun, Nov 30, 2008 at 15:31, Mircea Zahan <mz...@adaptive.ro> wrote:
>> On Sun, Nov 30, 2008 at 10:23, Mircea Zahan <mz...@adaptive.ro> wrote:
>>>
>>> Well now, for someone proficient with a programming language,
>>> recompiling sources for just another flavor of Windows  should be
>>> a task of what ... a few minutes ?
>>
>> Plus testing. Or do you think that person would be comfortable with
>> releasing binaries and then answering any/all bugs that arise with
>> "well, I didn't bother testing"? These aren't anonymous releases.
>
> I get your point, that did not cross my mind. But I myslef could live,
> for the time being, with some binaries compiled with "use it at your own
> risk"
> option :) At least that would be a start.

Unfortunately, it's too easy for people to ignore that and lay blame
on people who were "just trying to help" by making a premature release
that isn't tested enough.

>> Are you certain that there's nothing that would need to be tweaked in
>> the source code for a 64-bit environment?
>
> No, I am not. As I already said, I do not master C/C++. But from
> a Delphi programmer's point of view, compiling for x64 is just a
> matter of changing a compiler option. Maybe that's not the case
> with C/C++, but I can hardly imagine what would need to be
> tweaked except the target platform.

I think it would be wise to investigate the source & compilers before
making statements that "it should just be a compiler option."

>>
>> I don't know whether it's possible to compile Windows 64-bit binaries
>> on 32-bit Windows. If it's not, then it becomes a matter of cost as
>> well - if the volunteers who are doing this don't have 64-bit
>> hardware, they'll have to buy a system. And by extension, a copy of
>> 64-bit Windows too. So now it's a matter of  at least $500 for a
>> single-purpose machine to compile a few releases a year on a
>> *volunteer* basis.
>
> 64 hardware has emerged since more than 4 years. Nowadays one
> has to do some serious digging to find a 32 bit system on store shelfs. I
> doubt that a 64 bit hardware is an investment problem, as most people
> have already did that investment.

It doesn't take a lot of digging. I've been shopping for a new laptop
for my wife for a few months and have seen plenty shipping with 32-bit
processors. And even on 64-bit hardware, 32-bit Vista is frequently
shipped - again, requiring that one purchase an upgraded license to
get the 64-bit version.

>> Did 64-bit Windows change significantly between Windows XP x64 and
>> future releases (2003 Server, 2008 Server, Vista 64)? If so, which
>> releases do you actually support?
>
> We are talking about an application which uses TCP/IP and file access
> functions,
> not some exotic API. I'm pretty sure nothing has changed in that matter.
> Even more, the code is written (or it should be) using, if I remember well,
> the network/file access function from C libraries, which provide
> transparency,
> hidding API details and OS flavor differences.

But as you've noted many times, you're not a C/C++ coder, nor have you
looked at the source - so how do you know this for sure? And we're not
just talking about the Subversion source here - there's APR, zlib and
Neon at the very least.

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

Re: x86 is stone age, how about moving along ?

[Mircea Zahan <mz...@adaptive.ro>]

I'm not going to take this debate any further because
a debate surely wasn't my intent from the very beginning.
I think each of us are to some degree right and some
degree wrong and the truth is somewhere in the middle.

As a previous poster (Hilco Wijbenga) said,
I should try compiling x64 binaries myself. And I will.

Bottom line: no x64 binaries for geeks like me.

Kind regards to everyone,
Mircea.


----- Original Message ----- 
From: "Andy Levy" <an...@gmail.com>
To: "Mircea Zahan" <mz...@adaptive.ro>
Cc: <us...@subversion.tigris.org>
Sent: Monday, December 01, 2008 2:48 AM
Subject: Re: x86 is stone age, how about moving along ?


> On Sun, Nov 30, 2008 at 15:31, Mircea Zahan <mz...@adaptive.ro> wrote:
>>> On Sun, Nov 30, 2008 at 10:23, Mircea Zahan <mz...@adaptive.ro> wrote:
>>>>
>>>> Well now, for someone proficient with a programming language,
>>>> recompiling sources for just another flavor of Windows  should be
>>>> a task of what ... a few minutes ?
>>>
>>> Plus testing. Or do you think that person would be comfortable with
>>> releasing binaries and then answering any/all bugs that arise with
>>> "well, I didn't bother testing"? These aren't anonymous releases.
>>
>> I get your point, that did not cross my mind. But I myslef could live,
>> for the time being, with some binaries compiled with "use it at your own
>> risk"
>> option :) At least that would be a start.
> 
> Unfortunately, it's too easy for people to ignore that and lay blame
> on people who were "just trying to help" by making a premature release
> that isn't tested enough.
> 
>>> Are you certain that there's nothing that would need to be tweaked in
>>> the source code for a 64-bit environment?
>>
>> No, I am not. As I already said, I do not master C/C++. But from
>> a Delphi programmer's point of view, compiling for x64 is just a
>> matter of changing a compiler option. Maybe that's not the case
>> with C/C++, but I can hardly imagine what would need to be
>> tweaked except the target platform.
> 
> I think it would be wise to investigate the source & compilers before
> making statements that "it should just be a compiler option."
> 
>>>
>>> I don't know whether it's possible to compile Windows 64-bit binaries
>>> on 32-bit Windows. If it's not, then it becomes a matter of cost as
>>> well - if the volunteers who are doing this don't have 64-bit
>>> hardware, they'll have to buy a system. And by extension, a copy of
>>> 64-bit Windows too. So now it's a matter of  at least $500 for a
>>> single-purpose machine to compile a few releases a year on a
>>> *volunteer* basis.
>>
>> 64 hardware has emerged since more than 4 years. Nowadays one
>> has to do some serious digging to find a 32 bit system on store shelfs. I
>> doubt that a 64 bit hardware is an investment problem, as most people
>> have already did that investment.
> 
> It doesn't take a lot of digging. I've been shopping for a new laptop
> for my wife for a few months and have seen plenty shipping with 32-bit
> processors. And even on 64-bit hardware, 32-bit Vista is frequently
> shipped - again, requiring that one purchase an upgraded license to
> get the 64-bit version.
> 
>>> Did 64-bit Windows change significantly between Windows XP x64 and
>>> future releases (2003 Server, 2008 Server, Vista 64)? If so, which
>>> releases do you actually support?
>>
>> We are talking about an application which uses TCP/IP and file access
>> functions,
>> not some exotic API. I'm pretty sure nothing has changed in that matter.
>> Even more, the code is written (or it should be) using, if I remember well,
>> the network/file access function from C libraries, which provide
>> transparency,
>> hidding API details and OS flavor differences.
> 
> But as you've noted many times, you're not a C/C++ coder, nor have you
> looked at the source - so how do you know this for sure? And we're not
> just talking about the Subversion source here - there's APR, zlib and
> Neon at the very least.

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org