gnupic: Thread: gputils 1.0.0 Release Candidate 1

Dear gputils users,

gputils-1.0.0 Release Candidate 1 is available at 
http://sourceforge.net/projects/gputils/files/gputils/1.0.0/gputils-1.0.0_RC1.tar.gz.
Widows 32bit setup package is at 
http://sourceforge.net/projects/gputils/files/gputils-win32/1.0.0/gputils-1.0.0_RC1.exe.

This gputils 1.0.0 includes the last big missing functionality comparing 
with MPASM(X): support for nested WHILE loops. It also includes many 
improvements in listing file and error message generation. Some bugs 
were fixed, hopefully not many new introduced.

Please test the release candidates as much as possible and let me know 
about ANY results: pass or fail!

Final release is planned for end of year 2012, but it depends on number 
of critical bugs found.

If you find regressions or other bugs, please report them to the bug 
tracker at SourceForge. Also, feel free to give feedback here on the 
mailing list.

Borut

Borut Ražem wrote:
> gputils-1.0.0 Release Candidate 1 is available

Lovely!

A small idea: Skip the last version number digit and go for 1.0?


//Peter

Le lundi 10 décembre 2012 02:25:32 Peter Stuge a écrit :
> Borut Ražem wrote:
> > gputils-1.0.0 Release Candidate 1 is available
> 
> Lovely!
> 
> A small idea: Skip the last version number digit and go for 1.0?

Cleary, not!
I never heard a such bad idea...

foo-x.y.z is a rule accepted by all in the community for a very long time....

Why do you want to break it?

Alain

Alain Portal wrote:
> > A small idea: Skip the last version number digit and go for 1.0?
> 
> Cleary, not!
> I never heard a such bad idea...
> 
> foo-x.y.z is a rule accepted by all in the community for a very
> long time....
> 
> Why do you want to break it?

x, y and z have no purpose. They are universally only for show.

Microsoft realized that 18 years ago. (Windows 95)

Three feelgood for-show numbers is at least one more than gputils
actually needs, because development is moderate. That is *not* a
bad thing!

Why does anyone want to use an 8-bit machine when a 32-bit machine
can accomplish same task at nearly same cost with nearly same power
consumption?

The answer of course lies in the "nearly" - cut overhead where it
makes sense.


//Peter

Peter Stuge writes:
>Alain Portal wrote:
>> > A small idea: Skip the last version number digit and go for 1.0?
>> 
>> Cleary, not!
>> I never heard a such bad idea...
>> 
>> foo-x.y.z is a rule accepted by all in the community for a very
>> long time....
>> 
>> Why do you want to break it?
>
>x, y and z have no purpose. They are universally only for show.
>
>Microsoft realized that 18 years ago. (Windows 95)
>
>Three feelgood for-show numbers is at least one more than gputils
>actually needs, because development is moderate. That is *not* a
>bad thing!

They aren't feelgood for-show numbers when they actually express a
point in the development process -- which they do in this case.

>Why does anyone want to use an 8-bit machine when a 32-bit machine
>can accomplish same task at nearly same cost with nearly same power
>consumption?
>
>The answer of course lies in the "nearly" - cut overhead where it
>makes sense.
>
>
>//Peter
>
>---------------------------------------------------------------------
>To unsubscribe, e-mail: ####@####.####
>For additional commands, e-mail: ####@####.####
>

-- 
Joseph J. Pfeiffer, Jr., Ph.D.                 http://pfeifferfamily.net/
1440 Tierra del Sol Dr                         575.525.2764 (H)
Las Cruces, NM 88007                           575.496.3501 (C)

Le lundi 10 décembre 2012 04:18:25 Peter Stuge a écrit :
> Alain Portal wrote:
> > > A small idea: Skip the last version number digit and go for 1.0?
> > 
> > Cleary, not!
> > I never heard a such bad idea...
> > 
> > foo-x.y.z is a rule accepted by all in the community for a very
> > long time....
> > 
> > Why do you want to break it?
> 
> x, y and z have no purpose. They are universally only for show.

No purpose????
Only for show????

Sorry, I'm french, and my english is to bad that I can really express my anger 
on about such stupid word or thought... I apparently already struggling to 
make myself understood in French when I know I'm right, then in English ...
But I'll try ...
(translated with google: ( )

If the maintainer of the software forgets to integrate the license in the 1.0 
version (1.0.0), it should create a version 1.1 to fix this? Version 1.0.1 is 
sufficient for this ...

If the maintainer of the software decides to include a new translation of man 
pages in version 1.1 (1.0.1), it should create a version 1.2 for this? Version 
1.0.2 is sufficient for this ...

If the maintainer of the software realizes he made ​​a spelling mistake (we 
reported it) in version 1.2, it should create a version 1.3 for this? Version 
1.0.3 is really enough for this ...

In x.y.z, z is the minimum for the respect of those who report a problem ...

x.y.z is a commonly accepted rule in the Linux community, why do you ignore? 
Why do you want to change this rule so simple? Why do you try to impose a new 
model?
Why?
Why?

In version 1.1 (1.1.0), I expect something other than correcting a spelling 
mistake, the translation of the manual pages Lithuanian, and adding the 
license forgotten ...

z is much more useful than believed people who come to see once a year if 
there is a new version of their favorite software, because other come see 
every day ...

I do not even hope to be understood, especially by you :(

Those who have something other than pitch chick in their head will understand, 
even if they have not appreciated my way of expressing myself ...

Translation is fully copyrighted by translate.google  :(

Please...
Go back to Windows!

With no regards!
Alain

Alain Portal wrote:
> If the maintainer of the software forgets to integrate the license
> in the 1.0 version (1.0.0), it should create a version 1.1 to fix
> this?

Sure - why not?


> Version 1.0.1 is sufficient for this ...

Sure, just like 1.0.0.0.1 is sufficient. My point is that the number
of zeros between those two ones is quite arbitrary.


> If the maintainer of the software decides to include a new
> translation of man pages in version 1.1 (1.0.1), it should create
> a version 1.2 for this?

Again - why not?


> Version 1.0.2 is sufficient for this ...

Sure, just like 1.0.0.0.2 is sufficient. The number of zeros is
arbitrary.


> x.y.z is a commonly accepted rule in the Linux community,

No, not really. It is used by many packages, but many other packages
use other numbering. As you probably know the Linux kernel uses x.y
numbers since 3.0. Various packages even use only a single x number.


> why do you ignore?

I'm looking at what makes sense. For gputils I think it would make
sense to use one number less, because development is moderate. In
case that should change there's also no problem to add a number
later.


> Why do you want to change this rule so simple? Why do you try to
> impose a new model?
> Why?
> Why?

I actually mentioned why already in my first reply. I hope the above
helps clarify.


> In version 1.1 (1.1.0), I expect something other than correcting a
> spelling mistake, the translation of the manual pages Lithuanian,
> and adding the license forgotten ...

I'd suggest to read the changelog for each release, rather than to
make assumptions about what has changed based on the version numbers.


> z is much more useful than believed people who come to see once a year if 
> there is a new version of their favorite software, because other come see 
> every day ...

I'm not sure I see the difference. Regardless of how often, the
visitors who stop by to check for updates notice that there has
been a new release, or there has been no new release.

A few numbers can never convey what has changed.


> Please...
> Go back to Windows!

If you prefer, feel free to exchange the Windows 95 example with
Fedora Core 17 or Ubuntu 12.10.

Canonical went further and stripped the numbers of all significance,
making them represent a simple snapshot date. (year 2012 month 10)

That's fine for a project with significant development, where "lots"
can be expected to change between snapshots, but for gputils I think
1.0 is a nice sweet spot.

Any decision is of course for the maintainer to make, neither Alain
nor me. I just thought now was a good time to bring it up.


//Peter

Le 10/12/2012 05:43, Alain Portal a écrit :
> Please...
> Go back to Windows!
That's actually a pretty bad example, because I'm currently running 
explorer.exe-6.0.2900.5512, but never mind.

The long version strings are just related to an habit of using small 
version numbers. Or a huge development flow.

programs with 1.103 version numbers have no reason to be a problem.


Le 10/12/2012 05:43, Alain Portal a écrit :
> If the maintainer of the software decides to include a new translation of man
> pages in version 1.1 (1.0.1), it should create a version 1.2 for this? Version
> 1.0.2 is sufficient for this ...
When you say something like that, it's obvious you're motivated by 
feelings, not reason. If you accept large numbers, the position of a 
number in a string of dots does not indicate anything about the 
magnitude of the change! "Why not" is a perfecly valid answer to your 
question.

The only requirement I see in version strings is that they shall 
indicate a progression in a "lexical" order (1.0<1.1<2.0; 10.04<12.06). 
Apart from that, i don't see a strict requirement in the number of 
subparts in any version string.

Sebastien

> This gputils 1.0.0 includes the last big missing functionality comparing 
> with MPASM(X): support for nested WHILE loops. It also includes many 
> improvements in listing file and error message generation. Some bugs 
> were fixed, hopefully not many new introduced.

is it possible to compile BMinch's USB stack from 
http://pe.ece.olin.edu/ece/projects.html directly without any changes ?
Has anybody tried it ?

On 10. 12. 2012 09:07, Vaclav Peroutka wrote:
>> This gputils 1.0.0 includes the last big missing functionality comparing
>> with MPASM(X): support for nested WHILE loops. It also includes many
>> improvements in listing file and error message generation. Some bugs
>> were fixed, hopefully not many new introduced.
> is it possible to compile BMinch's USB stack from
> http://pe.ece.olin.edu/ece/projects.html directly without any changes ?
> Has anybody tried it ?

See http://sourceforge.net/p/gputils/bugs/188/: the attached 
lab2_18F2455.tgz compiles OK. I can't tell it for the original 
http://pe.ece.olin.edu/ece/projects/lab2_18F2455.zip since I can't reach 
the page and/or download the file.

Borut