Hi Seth,

I'm not opposed to the idea, however, I would like to make this code available quickly to halt the "exec_cmd" problem in slim_source itself, and going through the process of putting this in another consolidation is an effort I don't currently have time for. Perhaps I can move it down the road (or maybe you or someone you know would be interested in championing that cause) - in the meantime, it will be available in pkg:/system/library/install, which ends up on all systems (though admittedly a strange dependency to introduce into code).

- Keith

On 01/31/11 01:53 PM, Seth Goldberg wrote:
Hi,

As another Python author who needs to exec() stuff, I'll ask: Is there any way this can be put into a library that can be used by stuff outside of install? (i.e. into the pysolaris stuff in ON or some other general-purpose package that can be shared across consolidations)?

 --S

Quoting Keith Mitchell, who wrote the following on Mon, 31 Jan 2011:

On 01/31/11 01:29 PM, Drew Fisher wrote:
Trying this again since something caused the first one to bounce...


Keith,

__init__.py:

1 - do you need the shebang here?

Due to quirks in pkgdepend(1), yes, all Python files in our gate should have a Python shebang line. (If you know of any that don't, that's a bug)

150 - change to:    if logger is not None:
151 - remove this line?

Updated.


Other than that, this looks great. Where was this 7 months ago when we did DC? :)

7 months ago I was still convinced that subprocess.Popen was "good enough" for anything we had to do! Sorry!

- Keith


-Drew

On 1/31/11 1:20 PM, Keith Mitchell wrote:
Hi all,

I'd like a code review for my fixes for:

15957 <http://defect.opensolaris.org/bz/show_bug.cgi?id=15957> Implement solaris_install.Popen 7014402 <http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=7014402> SUNW* packages showing up in install-redistributable


webrev:
http://cr.opensolaris.org/~kemitche/webrev.15957/

Notes:
* solaris_install.Popen is a drop-in replacement for subprocess.Popen - as verification, the test code actually runs the full subprocess.Popen test suite against solaris_install.Popen. * "exec_cmd" and its family have not yet been removed. I will file bugs against the various installers to replace usage of such functions with solaris_install.Popen invocations; this allows for new consumers to use solaris_install.Popen now, rather than waiting an indefinite amount of time for me (or someone) to pull out those functions and widely re-test a large portion of the slim_source codebase. * I initially considered doing a global find/replace of subprocess.Popen with solaris_install.Popen. However, since solaris_install.Popen is in the solaris_install module, in many cases that results in a new Python import of a module not in the Python standard library. This has potential implications on packaging dependencies; as such, there are risk and time factors involved in such a find/replace (in the form of at least sanity checking most installers). As there will be bugs to remove "exec_cmd" functions anyway, it seems appropriate to have those bugs also replace subprocess.Popen with solaris_install.Popen. (The current approach also is probably easier for the various projects to merge with)

Thanks,
Keith


_______________________________________________
caiman-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/caiman-discuss


_______________________________________________
caiman-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/caiman-discuss



_______________________________________________
caiman-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/caiman-discuss

Reply via email to