Re: amake: a new build tool
Marco
host-address--so-tickle-me> wrote:
> Hi Folks,
>
> I am posting to publicize a free variant of make, named
> amake, that I developed. It is based on GNU Make. I've
> tested it on several x86 Linux distributions. It is
> available at:
>
> http://posixamake.sourceforge.net/
>
> This is its blurb:
>
> POSIX amake is a software-build tool, which adds automatic
> dependency analysis to GNU Make. It monitors and records
> commands executed, programs executed, shared libraries
> opened, files accessed, current working directory, and
> environment-variable values.
>
> Its documentation is in Chapter 13 of the modified GNU Make
> Manual, which is at:
>
> http://posixamake.sourceforge.net/amake.html/Amake-Features.html#Amak...
>
> If you are interested, please try it, and tell me what you
> think about it.
>
> By the way, I did not post to comp.software.config-mgmt,
> instead, because I have not seen posts there in over a week.
>
> -- Jim
hmm another sourceforge "amake" project also exists
Jim Buffenbarger
Jim> I am posting to publicize a free variant of make, named
Jim> amake, that I developed. It is based on GNU Make. I've
Jim> tested it on several x86 Linux distributions. �It is
Jim> available at:
Jim>
Jim> � �http://posixamake.sourceforge.net/
Marco> hmm another sourceforge "amake" project also exists
Marco>
Marco> http://amake.sourceforge.net/
Yes, that's true. Perhaps I should change its name. I chose
the name "amake" before I started using sourceforge for
hosting.
The amake you mention seems to combine features found in
other versions of make. My amake tries to add my favorite
features of Clearcase and Vesta, which are more than just
versions of make.
-- Jim
Marc Girod
host-address--so-tickle-me> wrote:
> My amake tries to add my favorite features of Clearcase
> and Vesta, which are more than just versions of make.
Thus 'a' stands for 'audited'?
Excellent.
Do you also offer or plan to offer winkin?
Have a repository to save the derived objects?
Are you aware of David Boyce's AO
(for Audited Objects):
Marc
Jim Buffenbarger
Jim> My amake tries to add my favorite features of Clearcase
Jim> and Vesta, which are more than just versions of make.
Marc> Thus 'a' stands for 'audited'? Excellent.
My initial intent was 'a' for 'access', since it monitors
file accesses, via a shared library named libaccess.so. I
like your 'audited', since it also monitors the commands
that are executed. I suppose 'automatic' makes some sense,
too.
Marc> Do you also offer or plan to offer winkin? Have a
Marc> repository to save the derived objects?
I have certainly thought about those features, and will
continue to do so.
As you probably know, Vesta's ability in those areas is a
generation beyond what Clearcase offers.
Marc> Are you aware of David Boyce's AO (for Audited Objects):
Marc>
Marc> http://clearguide.com/
Now I am. I'll investigate. Thanks!
-- Jim
Marc Girod
host-address--so-tickle-me> wrote:
> As you probably know, Vesta's ability in those areas is a
> generation beyond what Clearcase offers.
No, I didn't...
It's already some time I haven't check Vesta.
I was blocked at a time by the lack of support for Windows.
Also, it required that all the tools would be stored in the Vesta
repository,
IIRC...?
ClearCase has practically stopped being developed 15 years ago
in these areas... So no surprise.
In what way has however Vesta made progress?
The reason I am a great believer in auditing is the possibility
to turn SCM upside down, by making derived objects the
prototypical configuration item.
This is made possible by avoiding duplication.
Thus build avoidance and winkin are the first step.
Next, one would like lazy identification, i.e. the possibility
to tag different files as instances of a same entity.
And especially over multisite: allow you to compare
objects built independently on different hosts, e.g. one
built locally, and the other downloaded.
This kind of situation is met with tools like maven and
Buckminster.
Marc
Jim Buffenbarger
Jim> As you probably know, Vesta's ability in those areas is a
Jim> generation beyond what Clearcase offers.
Marc> No, I didn't...
Marc> It's already some time I haven't check Vesta.
Marc> I was blocked at a time by the lack of support for Windows.
Marc> Also, it required that all the tools would be stored in the Vesta
Marc> repository,
Marc> IIRC...?
Yes, those aspects of Vesta have not changed.
Personally, I like the tools-in-the-repository requirement,
but I can understand why some people wouldn't like it.
Marc> ClearCase has practically stopped being developed 15 years ago
Marc> in these areas... So no surprise.
That is also true.
Marc> In what way has however Vesta made progress?
Aside from incremental improvements, the fact that it runs
on Linux distributions is important to me.
Marc> The reason I am a great believer in auditing is the possibility
Marc> to turn SCM upside down, by making derived objects the
Marc> prototypical configuration item.
Marc> This is made possible by avoiding duplication.
Marc> Thus build avoidance and winkin are the first step.
I agree.
Clearcase requires winkin of fine-grained derived objects
before winkin of dependent large-grained derived objects
(e.g., it must winkin an object file before winking-in an
executable file). Vesta allows the immediate "winkin" of a
large-grained derived object, which saves much time and space.
Marc> Next, one would like lazy identification, i.e. the possibility
Marc> to tag different files as instances of a same entity.
Yes, this is important. With Clearcase, "interchangable"
fine-grained derived objects can cause unnecessary creation
of functionally identical large-grained derived objects.
Marc> And especially over multisite: allow you to compare
Marc> objects built independently on different hosts, e.g. one
Marc> built locally, and the other downloaded.
Marc> This kind of situation is met with tools like maven and
Marc> Buckminster.
Neither Clearcase or Vesta allow this. Clearcase assumes
slow remote access. Vesta assumes unreliable remote access.
-- Jim