[LEAPSECS] POSIX and C (Was: Re: ISO Influence)
seaman at noao.edu
Sun Dec 19 10:44:25 EST 2010
On Dec 19, 2010, at 2:07 AM, Poul-Henning Kamp wrote:
> In message <1292742460.31540.137.camel at localhost>, Paul Sheer writes:
>> For this and other reasons many programs implement their own function to
>> do this. They would assume 86400 seconds per day to copy POSIX or
>> because they have never heard of leap seconds. I would guess there are
>> an enormous number of statically linked executables that contain such
> Well, enourmous or not, we simply do not know, and Rob doesn't think
> it is so important to find them, that we should develop a time-estimate
> for how long time it will take.
Oh puh-lease. Rob rather thinks this is the responsibility of those advocating a drastic change to an underlying standard.
I'm sure many of those on the list were involved in Y2K remediation and could comment on the effort needed to perform such an inventory. Demanding an estimate in "lines of code per hour" is disingenuous. A human obviously did not scrutinize every line of code, rather pattern matching was performed on tell-tale tokens like "1900", "century" and "year". The precise pattern matching expression undoubtedly varied from project to project, for instance astronomers looking for Y2K problems would have searched on "DATE-OBS". The expressions were tweaked in response to problem areas as they were detected and understood in context. Etc and so forth.
In addition, someone familiar with a particular code hierarchy performed a top-down functional review of the interfaces. At the National Observatory, I was tasked with supervising these efforts (the whole team participated) for our big community-wide image processing package. Folks in another group handled the telescope control systems. Numerous folks on staff and in the community reviewed the software associated with our dozens of cameras. The astronomical community worked together in the most gratifying way to revise our common data standard.
Worldwide, Y2K-related efforts likely cost the astronomy community between 10 and 100 person-years. No idea what the capital costs might have been for accelerating server upgrades, etc. I would not be surprised if costs on the business side (financial and administrative computing, etc) were larger yet. By all means feel free to turn this back into units of lines of code per hour if it floats your boat.
Absolutely none of this has been done for leap seconds. In astronomy we did perform an initial review when these leap second shenanigans first came to our attention. This review was enough to show that the proposed change would cause our community great expense and introduce significant risks. Thus (and for other more fundamental reasons) we oppose the change.
Advocates of redefining UTC say that astronomers are the only ones who will be affected. Nobody, however, has invested the least effort in finding out if this self-serving assumption is true.
It is a very strong argument for continuing the status quo to assert that it is too expensive to even perform the study to identify the costs and risks of introducing the proposed change.
The ITU proposal is ridiculously incomplete and amateurish. It should be rejected out of hand even by conscientious proponents of a change to UTC.
More information about the LEAPSECS