[oh-tracker] [ openhpi-Bugs-3457592 ] oh_decode_time() decodes relative times as "invalid time"

[SourceForge.net <no...@so...>]

Bugs item #3457592, was opened at 2011-12-11 14:40
Message generated for change (Comment added) made by mwalle
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=532251&aid=3457592&group_id=71730

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Michael Walle (mwalle)
Assigned to: Nobody/Anonymous (nobody)
Summary: oh_decode_time() decodes relative times as "invalid time"

Initial Comment:
SaHpi.h defines SAHPI_TIME_UNSPECIFIED as the lowest possible number (remember SaHpiTimeT is signed).

oh_decode_time() checks for (time > SAHPI_TIME_UNSPECIFIED) which always evaluates to true. Therefore, the else branch will never be executed.

There should never be an "Invalid time" if relative timestamps can be negative.

----------------------------------------------------------------------

Comment By: Michael Walle (mwalle)
Date: 2011-12-12 05:09

Message:
yeah, but 8.1 states:
When reporting a relative time, the specific meaning of "startup" is
implementation dependent. [..]

This section always uses the term "startup", thats the reason for my
suggestion.

OTOH you could use custom format field and let the user of oh_decode_time()
decide what to use. Of course there would have to be two formats, one for
relative and one for absolute times and unless you create a new function
we'll break backwards compatibility ;)

sth like:
oh_decode_time_fmt(const char *fmt1, const char *fmt2, ...)
oh_decode_time() which interally uses oh_decode_time_fmt()

----------------------------------------------------------------------

Comment By: Anton Pak (avpak)
Date: 2011-12-12 04:33

Message:
Section 6.4.6 says:
If the implementation cannot supply an absolute time value, then it may
supply a time relative to some system-defined epoch, such as system
startup.

So startup is only one of the possible options.

----------------------------------------------------------------------

Comment By: Michael Walle (mwalle)
Date: 2011-12-12 04:30

Message:
Section 8.1 specifies relative times 'since startup'. So there should be a
check for negative relative times.

I suggest an output like:
 "[[[<d> days] <h> hours] <m> minutes] <s> seconds since startup"
or if that is too long use d, h, m, s for days, hours, minutes, seconds
rsp.

----------------------------------------------------------------------

Comment By: Anton Pak (avpak)
Date: 2011-12-12 03:06

Message:
Yes, the check for (time > SAHPI_TIME_UNSPECIFIED) is incorrect.
And probably the output for relative time is incorrect too.
As I can see we can have up to ~10000 days for max relative time.
What do you think is the best way to make textual representation of the
relative time?
What about "REL <x> seconds" or "REL <x> hours <y> seconds"?


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=532251&aid=3457592&group_id=71730