Latest SWIG, SQLite3 bindings for ruby, Python PCSI kit broken?

[Ben Armstrong]

Buoyed by my partial success with SQLite3, I'm now looking at what it would take to get the ruby bindings working. Now, they require SWIG (http://www.swig.org/) which does have VMS support in the package available from upstream. Unfortunately, the DCL build procedure shipping with SWIG is out of date and the script to update the DCL procedures is written in Python. So I went to JFP's site to download and install his latest Python kit on our OpenVMS 8.3 Alpha test system:

JFP-AXPVMS-PYTHON252-V0100-0-1.PCSI

Unfortunately, it does not install. It appears to start out well, but after watching the progress indicator tally percentages up to 100%, I get the following error:

$ product install python252
...
The following product will be installed to destination:
JFP AXPVMS PYTHON252 V1.0-0 DISK$ALPVMSSYS:[VMS$COMMON.]
...

Now, I don't really need Python at this time so it's not crucial that we solve this problem. All I really want are the updated .com files in the [.vms] directory of SWIG 1.3.35. Could someone with a working Python please provide them?

Thanks,
Ben

[Ben Armstrong]

Oops. I guess I was distracted while waiting for the Python re-install attempt so I could cut-and-paste the error message into this post so that I ended up posting without it. :)

Here's the actual error:

Portion done: 0%...10%...20%...30%...40%...50%...60%...70%...80%...90%
%DCL-W-INSFPRM, missing command parameters - supply all required parameters
%RENAME-E-SEARCHFAIL, error searching for PCSI_DESTINATION:[000000]python252.DIR;1
-RMS-F-DEV, error in device name or inappropriate device type for operation
%PCSI-I-SPAWNEXE, error executing: @PCSI$SOURCE:[TMP]PYTHON252$POST_CONFIGURE.COM

%PCSI-E-EXEPSTFAIL, product supplied EXECUTE POSTINSTALL procedure failed
-RENAME-E-SEARCHFAIL, error searching for !AS
%PCSI-E-OPFAILED, operation failed
Terminating is strongly recommended. Do you want to terminate? [YES]

[Jean-François Piéronne]

Ben,

The simple way to install Python is to use the 2 LD images.

More information on
http://vmspython.dyndns.org/DownloadAndInstallationPython


Are your sure to have installed the latest pcsi kit, because this kit has already been installed without any problem.

Anyway I can provide a build version of SWIG 1.3.31 which is the version I use for building some Python extension.

JF

[Ben Armstrong]

Sure. If you could provide Swig 1.3.31 that would be great. If I revisit python later, I'll look into the LD images, but that's quite a lot more stuff than we really need, and this is a small test system, so I didn't want to install anything more than I need at this time.

I'll also check PCSI, but I thought we upgraded that recently.

Thanks,
Ben

[Jean-François Piéronne]

Ok, I will put online a pre-build swig version.

About Python, do you have installed run the startup procedure for the zlib and libz2 libraries?

[Jean-François Piéronne · Accepted solution]

Download link:
http://vmspython.dyndns.org/anonymous/jfp/swig1331.zip

contains the axp version swig.exe

JF

[Ben Armstrong]

Yes, I installed Python's dependencies first, as per your doc. Thanks for SWIG. Downloading and trying it now ...

[Jean-François Piéronne]

Ben,

do you use for the installation thr qualifier /destination with a logical name? this may be the problem.
If you have used /destination you can try to specified a full pathname (like dev:[dir]) where dev is a real device name.

JF

[Ben Armstrong]

No, I just used the default, as per the command shown above in my original post.

[Jean-François Piéronne]

The offending lines are:

$ def pcsi_destination/trans=conc 'f$trnlnm("PCSI$DESTINATION")'
$ set default pcsi_destination:[000000]
$ rename [000000]python252.DIR;1 [000000]python252.DIR;1

So it seem that pcsi_destination defined using PCSI$DESTINATION was wrong, but this works on all the systems I have tested.

Do you have installed any PCSI patch
on my system VMS732_PCSI V3.0 latest V5.0

JF

[Ben Armstrong]

DEC AXPVMS VMS83A_PCSI V1.0

[Hoff]

Recent PCSI kits -- since revised and replaced -- deleted part of PCSI itself.

Get the next PCSI kit, read the release notes, and follow the directions -- you'll probably need to recover a handful of PCSI*.COM files from your V8.3 distribution, and place them into the SYS$COMMON:[SYSUPD] area, IIRC.

These errant PCSI kits can derail JFP's MySQL kit installation, too.

Getting these and various other tools installed and running on OpenVMS -- which is a non-trivial undertaking, and which usually requires some local porting -- is a prerequisite toward incrementally increasing the portability of the associated local application code, too.

[Ben Armstrong]

After installing PCSI V2.0 for OpenVMS Alpha V8.3, another attempt to install python using the same command as in the original post failed. The messages are as follows:

...
The following product will be installed to destination:
JFP AXPVMS PYTHON252 V1.0-0 DISK$ALPVMSSYS:[VMS$COMMON.]

Portion done: 0%...10%...20%...30%...40%...50%...60%...70%...80%...90%
%DCL-W-INSFPRM, missing command parameters - supply all required parameters
%RENAME-E-SEARCHFAIL, error searching for PCSI_DESTINATION:[000000]python252.DIR;1
-RMS-F-DEV, error in device name or inappropriate device type for operation
%PCSI-I-SPAWNEXE, error executing: @PCSI$SOURCE:[TMP]PYTHON252$POST_CONFIGURE.COM

%PCSI-E-EXEPSTFAIL, product supplied EXECUTE POSTINSTALL procedure failed
-RENAME-E-SEARCHFAIL, error searching for !AS
%PCSI-E-OPFAILED, operation failed
Terminating is strongly recommended. Do you want to terminate? [YES]
%PCSI-E-CANCEL_WIP, termination resulted in an incomplete modification to the system
%PCSI-E-S_OPCAN, operation cancelled by request

Recovery pass starting ...
... etc.

[Hoff]

Please confirm that you have verified that the half-dozen files mentioned in the PCSI V2 ECO kit present and accounted for.

I haven't yet downloaded the V2.0 ECO kit and pulled it apparent to verify if the kit fixed the original DELETE bug, or if it fixed the original DELETE bug *and* recovered the files that were DELETEd by the original bug. (And if the files were restored by the kit, I'd probably go the extra step and DIFFERENCES the restored files against the files from the V8.3 distribution.)

In any event, pull the HP PCSI kit and JFP's PCSI kit apart yourself and go look at what each is doing here. If you're headed down this path and seeking some degree of expediency, you're likely going to want and even need to know how to do that. See HELP PRODUCT EXTRACT, et al. The DCL and PCSI source files involved here are fully recoverable and readable, after all.

[Ben Armstrong]

Yes, those files are there:

Directory SYS$COMMON:[SYSUPD]

PCSI$CREATE_ACCOUNT.COM;1
39 22-JUN-2007 18:29:19.27
PCSI$CREATE_NETWORK_OBJECT.COM;1
37 22-JUN-2007 18:29:21.33
PCSI$CREATE_RIGHTS_IDENTIFIER.COM;1
14 22-JUN-2007 18:29:22.58
PCSI$DELETE_ACCOUNT.COM;1
16 22-JUN-2007 18:29:23.54
PCSI$DELETE_NETWORK_OBJECT.COM;1
23 22-JUN-2007 18:39:01.91
PCSI$DELETE_RIGHTS_IDENTIFIER.COM;1
13 22-JUN-2007 18:29:25.33


In the absence of evidence to the contrary, I'll have to assume the PCSI patch kit is good and the Python kit still has problems. While I'm interested to know if PCSI is working or not, JFP has already provided me with a pre-built version of SWIG, so that takes the pressure off me getting Python working. Except for building SWIG, I have so far managed to live without Python completely. This makes it unlikely that in the near term I'll be able to do the debugging you suggest.

Thanks,
Ben

[Shi Pu]

Is there any update? I met the same failure, on OpenVMS8.2 Alpha.

[labadie_1]

>>> Is there any update? I met the same failure, on OpenVMS8.2 Alpha.

This is a kit for Alpha Openvms 8.3, it is normal that it fails with 8.2

By the way, the preferred way for using Python is the LD devices, see
http://vmspython.dyndns.org/DownloadAndInstallationPython

[Jean-François Piéronne]

> This is a kit for Alpha Openvms 8.3, it is normal that it fails with 8.2

No, the AXP kit is for 7.3-2 (+patches)
It's the IA64 kit witch is for 8.3.

What is the exact error? I was unable to reproduce any installation problem on all system I have tested.