Yo Hal!
On Wed, 15 Mar 2017 23:21:59 -0700
Hal Murray wrote:
> g...@rellim.com said:
> > I like checks for things that can never happen. Somehow many of
> > them eventually do get tripped, then the error is more obvious,
> > assuming a good error message.
>
> I do too, but there can be too
g...@rellim.com said:
> I like checks for things that can never happen. Somehow many of them
> eventually do get tripped, then the error is more obvious, assuming a good
> error message.
I do too, but there can be too much of a good thing. There is a real talent
for putting in the right check
Yo Hal!
On Wed, 15 Mar 2017 14:32:50 -0700
Hal Murray wrote:
> g...@rellim.com said:
> > I'm gonna go with this for now, it seems pretty hard to
> > misunderstand: #define NS_PER_S 10
> > #define S_PER_NS 1e-9
> > DItto for other units.
>
> Looks good to me.
I'm half done with the co
g...@rellim.com said:
> I'm gonna go with this for now, it seems pretty hard to misunderstand:
> #define NS_PER_S 10
> #define S_PER_NS 1e-9
> DItto for other units.
Looks good to me.
An alternative hack is to undef NANOSECOND in the top of timespecops.h
#include "ntp_calendar.h"
#inc
Yo Hal!
I'm gonna go with this for now, it seems pretty hard to misunderstand:
#define NS_PER_S 10
#define S_PER_NS 1e-9
DItto for other units.
I see no preactical reason for a fefine of 1e9 as NS_PER_S will
do fine for that too.
RGDS
GARY
--
Gary E. Miller writes:
>> They don't. Integer and floating point constants are different.
>
> But in this case the defines are the reciprocals of each other. Very
> bad collision.
I was responding to Hal's apparent assertion that NANOSECONDS defined as
10 and 1.e9 were the same number.
Yo Achim!
On Wed, 15 Mar 2017 19:31:43 +0100
Achim Gratz wrote:
> Hal Murray writes:
> > They all duplicate stuff already defined in timespecops.h
>
> They don't. Integer and floating point constants are different.
But in this case the defines are the reciprocals of each other. Very
bad co
Yo Hal!
On Wed, 15 Mar 2017 11:50:30 -0700
Hal Murray wrote:
> You can remove that out of range check. ctl_pitfs is only called
> twice to print the time of the last leap second and the expiration
> time of the file. They should be sane. We can add a sanity check but
> there is already a check
e...@thyrsus.com said:
>> I never figured out what the fs and ls were trying to tell me. Before the
>> cleanup, the fs were probably converting ntp l_fp epoch times.
> No, I don't think I changed those. Lost in the mists of time...
I was referring to my recent leap second cleanup.
> Thanks.
Hal Murray writes:
> They all duplicate stuff already defined in timespecops.h
They don't. Integer and floating point constants are different.
> I assume the compiler doesn't complain about redefines if the result doesn't
> change.
The gcc preprocessor does it that way, yes. It's considered a
Gary E. Miller writes:
> So, any preferences for different and dsescription names that
> do not conflict with the misleading *BSD names?
Use a namespace specific to NTP(sec) for all defines. *ducks*
Regards,
Achim.
--
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+
Factory
Hal Murray :
> A cleanup opportunity.
>
> fstostr in ntp_util only called from within ntp_util but defined in ntpd.h
> %04d%02d%02d%02d%02d
>
> lstostr in ntp_leapsec only called from within ntp_leapsec
> %04d-%02d-%02dT%02d:%02dZ
>
> ctl_putfs in ntp_control only called from within ntp_cont
Hal Murray :
>
> ./waf configure --out=xxx
> saves the xxx someplace. Following waf builds or checks use that directory.
>
> I'd like to have more than one directory in-progress at the same time and
> switch to the desired one. Is that possible? How?
I don't know how to do this. It is pos
On Wed, Mar 15, 2017 at 10:04 AM, Eric S. Raymond wrote:
> It is now. Also I've broken up the main sequence into daemon, tools, and
> ntpviz components. If you want to add options to suppress any of these,
> feel free.
This makes me happy. Gary and I had this IRC discussion yesterday.
On a f
Hal Murray :
> The old text in INSTALL had comments about something being needed at run time
> or only at build tiime. That information is no longer in buildprep.
It is now. Also I've broken up the main sequence into daemon, tools, and
ntpviz components. If you want to add options to suppress
15 matches
Mail list logo
