0001101: gupdatedb mktemp and !

(0005852)
car (reporter)
2009-04-05 05:16

This issue predates my adoption of the package, which doesn't excuse me for not checking that there were no open issues when I took it on. I know about it now, though.

(0006392)
car (reporter)
2009-07-04 01:16

It seems to me that the slightly harder route, to fix the script so that it *is* Bourne shell compatible, is probably the better fix.

At a brief glance, the problem seems to be a couple of if tests unadorned by []. Fixing these certainly gets rid of the warnings and the script appears to function properly afterwards. I'll double-check this later, submit a new package to testing, then submit a patch upstream if this fixes the issue.

(0009048)
jay (reporter)
2011-05-14 15:05

! is required by POSIX. See section 2.9.2 of http://pubs.opengroup.org/onlinepubs/009695399/utilities/xcu_chap02.html [^]

It looks like the ultimate problem is that the source distribution hard-codes the interpreter:
$ head -n1 locate/updatedb.sh
#! /bin/sh

In other words, findutils assumes /bin/sh is POSIX-compliant. If that's not true on your systems, you will need to path that I think.

Lack of mktemp is trickier. Looks like you may need your findutils package to depend on something that provides it, if the base OS does not.

(0009056)
james (reporter)
2011-05-16 10:48
edited on: 2011-05-16 10:56

The date on the POSIX reference is 2004 and this bug was registered in 2005. It is highly likely at that time, using Solaris 8, /bin/sh was not POSIX compliment. Whether it was or not is not relevant to the fact of the matter being that gupdatedb failed on a target system and was fixable using the suggestion I made in 2005.

On Solaris 9, the current base system, gupdatedb appear to works. Is this proof that if reports are ignored for long enough there exists a probability that bugs will fix themselves? And if we wait long enough this bug report could morph into the complete works of Shakespeare.

(0009175)
bwalton (administrator)
2011-07-02 19:32

Closing this since it only affects Solaris 8 which is not supported any longer.

-Ben

(0009176)
james (reporter)
2011-07-02 20:15

The mktemp errors still applies?

(0009177)
bwalton (administrator)
2011-07-03 03:44

Yes it does. Sorry. I've updated the build recipe and will push an update soon.

Thanks
-Ben

(0009178)
james (reporter)
2011-07-03 09:44

Thanks.

Notes
(0005852) car (reporter) 2009-04-05 05:16	This issue predates my adoption of the package, which doesn't excuse me for not checking that there were no open issues when I took it on. I know about it now, though.

(0006392) car (reporter) 2009-07-04 01:16	It seems to me that the slightly harder route, to fix the script so that it is Bourne shell compatible, is probably the better fix. At a brief glance, the problem seems to be a couple of if tests unadorned by []. Fixing these certainly gets rid of the warnings and the script appears to function properly afterwards. I'll double-check this later, submit a new package to testing, then submit a patch upstream if this fixes the issue.

(0009048) jay (reporter) 2011-05-14 15:05	! is required by POSIX. See section 2.9.2 of http://pubs.opengroup.org/onlinepubs/009695399/utilities/xcu_chap02.html [^] It looks like the ultimate problem is that the source distribution hard-codes the interpreter: $ head -n1 locate/updatedb.sh #! /bin/sh In other words, findutils assumes /bin/sh is POSIX-compliant. If that's not true on your systems, you will need to path that I think. Lack of mktemp is trickier. Looks like you may need your findutils package to depend on something that provides it, if the base OS does not.

(0009056) james (reporter) 2011-05-16 10:48 edited on: 2011-05-16 10:56	The date on the POSIX reference is 2004 and this bug was registered in 2005. It is highly likely at that time, using Solaris 8, /bin/sh was not POSIX compliment. Whether it was or not is not relevant to the fact of the matter being that gupdatedb failed on a target system and was fixable using the suggestion I made in 2005. On Solaris 9, the current base system, gupdatedb appear to works. Is this proof that if reports are ignored for long enough there exists a probability that bugs will fix themselves? And if we wait long enough this bug report could morph into the complete works of Shakespeare.

(0009175) bwalton (administrator) 2011-07-02 19:32	Closing this since it only affects Solaris 8 which is not supported any longer. -Ben

(0009176) james (reporter) 2011-07-02 20:15	The mktemp errors still applies?

(0009177) bwalton (administrator) 2011-07-03 03:44	Yes it does. Sorry. I've updated the build recipe and will push an update soon. Thanks -Ben

(0009178) james (reporter) 2011-07-03 09:44	Thanks.

Relationships

OpenCSW Bug Tracker