|
Re^13: ActiveState Perl 10.x and Crypt::SSLeay not installed errors on Windows 2003
by Gavin (Archbishop) on Mar 29, 2008 at 12:48 UTC
|
It installed successfully for me too on Windows XP ActivePerl 5.8 for what its worth!
UPDATE
But not in 5.10 | [reply] |
Re^13: ActiveState Perl 10.x and Crypt::SSLeay not installed errors on Windows 2003
by syphilis (Archbishop) on Mar 29, 2008 at 13:04 UTC
|
So what am I doing wrong???
Nothing that I can see. I'm seeing the same ppd file as you, and I've got the same version of PPM (4.01).
One difference is that when I run that config list command in the ppm shell, my output is simply arch = MSWin32-x86-multi-thread-5.10 . Could the problem lie with all that additional "gui.*" gunk that you're getting ?
(I, too, like to avoid getting closely involved with PPM.)
Cheers,
Rob | [reply]
[d/l]
[select] |
|
|
Could the problem lie with all that additional "gui.*" gunk that you're getting ?
Not as far as I can tell. That's just because I typed "ppm" at a command prompt once and it started the gui. I quit straight away, but I guess it stored some config somewhere in the interim.
I'm almost certain the problems arise because I have two versions of active state installed. 5.8.6 as my main installation, and 5.10.0 whilst I work out how to build all the things I have in the 5.8.6.
I speculate that when you install AS builds, they secrete away some config in the registry somewhere and then go fetch that to decide what build/architecture you have. And even though I being very judicious in ensuring that nothing in my envoronment points to the 5.8.6 installation before running PPM, their script is fetching something from the registry somewhere and making the decision based upon that.
If people would add a bit of trace information, __FILE__ & __LINE__ when issuing error messages, or at least some useful information like say:
ppm install failed: The PPD does not provide code to install for this
+platform [MSWin32-xyz-pqr]
tracking down and solving these problems would be so much simpler.
Which is exactly why I stopped using PPM3. And that was before they complicated the existing issues with the great bob ball of (OO) string by adding the gui.
Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
"Science is about questioning the status quo. Questioning authority".
In the absence of evidence, opinion is indistinguishable from prejudice.
| [reply]
[d/l] |
|
|
c:\Perl510\bin>
perl -MActivePerl::PPM::Client
-wle"print ActivePerl::PPM::Client->new()->arch"
MSWin32-x86-multi-thread-118.53
Now from where do you think it came up with that?
Because I've searched my entire bloody filesystem and registry and that string doesn't seem to appear anywhere! Gah!
If Gavin hadn't just installed 1002 and encountered the same failure, I might assume that the reason 1002 was released so soon after 1000, was to correct this...but that still doesn't explain where it comes from. The absence of the string anywhere on my system suggests it must be constructed at runtime. Why? Does it change from run to run?
Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
"Science is about questioning the status quo. Questioning authority".
In the absence of evidence, opinion is indistinguishable from prejudice.
| [reply]
[d/l] |
|
|
| [reply] |
|
|
$arch .= sprintf "-%vd", substr($^V, 0, 2) if $] >= 5.008;
Which produces:
C:\Perl510>perl -wle"print $]; print $^V; printf qq[-%vd], substr($^V,
+ 0, 2)"
5.010000
v5.10.0
-118.53
A few further questions arise:
- What is that sprintf trying to do? (What is pattern %vd besides the obvious:)
- How come other some other peoples copies aren't afflicted?
- Can anyone confirm their version of AS1000/1002 has the above line?
Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
"Science is about questioning the status quo. Questioning authority".
In the absence of evidence, opinion is indistinguishable from prejudice.
| [reply]
[d/l]
[select] |
|
|
|
|
|
|
|
|
|
|
Could the problem lie with all that additional "gui.*" gunk that you're getting ?
Not as far as I can tell. That's just because I typed "ppm" at a command prompt once and it started the gui. I quit straight away, but I guess it stored some config somewhere in the interim.
I'm almost certain the problems arise because I have two versions of active state installed. 5.8.6 as my main installation, and 5.10.0 whilst I work out how to build all the things I have in the 5.8.6.
I speculate that when you install AS builds, they secrete away some config in the registry somewhere and then go fetch that to decide what build/architecture you have. And even though I'm being very judicious in ensuring that nothing in my environment points to the 5.8.6 installation before running PPM, their script is fetching something from the registry somewhere and making the decision based upon that.
If people would add a bit of trace information, __FILE__ & __LINE__ when issuing error messages, or at least some useful information like say:
ppm install failed: The PPD does not provide code to install for this
+platform [MSWin32-xyz-pqr]
tracking down and solving these problems would be so much simpler.
Which is exactly why I stopped using PPM3. And that was before they complicated the existing issues with the great big ball of (OO) string by adding the gui.
Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
"Science is about questioning the status quo. Questioning authority".
In the absence of evidence, opinion is indistinguishable from prejudice.
| [reply]
[d/l] |
|
|
I speculate that when you install AS builds, they secrete away some config in the registry
I think that's possibly the case with the msi packages, but I don't think you'll strike such an issue if you install from the zip packages.
I have 3 ActiveState builds on this box (806, 822 and 1002), all of which were installed from the zip packages, and I don't get any grief from that. I think the msi packages offer an "uninstall" capability - and I think that dictates that the registry is going to be messed with. The zip packages don't have an uninstall capability and don't (afaik) screw with the registry in any way - even if one does let the installation process modify the path (which I don't allow). I personally wouldn't recommend the msi packages under *any* circumstances.
Cheers,
Rob
| [reply] |
|
|
Re^13: ActiveState Perl 10.x and Crypt::SSLeay not installed errors on Windows 2003
by Not_a_Number (Prior) on Mar 29, 2008 at 13:00 UTC
|
Well I get the same output as you from the lwp-request, but a slightly different result from ppm config list:
arch = MSWin32-x86-multi-thread-5.10
gui.filter =
gui.filter.fields = name abstract
gui.filter.type = installed
gui.geometry = 568x535+154+154
gui.install_area = site
gui.view.abstract = 1
gui.view.area = 1
gui.view.author = 0
gui.view.available = 1
gui.view.installed = 1
gui.view.name = 1
gui.view.repo = 0
gui.view.sortcolumn = name
gui.view.sortorder = -increasing
gui.view.statusbar = 1
gui.view.toolbar = 1
But whether that helps or not is way over my head, I'm afraid. :-(
Update:Hey, I just reread the OP title: "...errors on Windows 2003". Like Gavin, I'm using XP. Is that it? | [reply]
[d/l]
[select] |
|
|
| [reply] |
Re^13: ActiveState Perl 10.x and Crypt::SSLeay not installed errors on Windows 2003
by Anonymous Monk on Mar 30, 2008 at 12:29 UTC
|
| [reply] |
|
|
ppm had that bug for 5.8
Except that, on 5.8, it wasn't a bug:
C:\>perl -e "$arch .= sprintf \"-%vd\", substr($^V, 0, 2) if $] >= 5.0
+08;print $arch"
-5.8
It wasn't until 5.10 that it became a bug:
C:\>perl -e "$arch .= sprintf \"-%vd\", substr($^V, 0, 2) if $] >= 5.0
+08;print $arch"
-118.53
Cheers,
Rob | [reply]
[d/l]
[select] |
|
|
$arch was wrong, now its wrong again
| [reply] |