tomc@oakhill.UUCP (Tom Cunningham) (07/15/87)
I noticed that in the latest release of Sun 3 software which we have (I
believe it is 3.2), there is no -p option for lint to perform portability
checks. This option does exist, however, in the System V version of lint
presumably distributed with the Sun 3.2 software.
Did Berkeley lint (if there is such a thing) ever have this option, was
it excised for some reason, or what? I am just curious, as I consider
the constraints of eight character symbol names (six characters external)
as somewhat of an anachronism portability-wise, but I could well be wrong.
Would anyone care to comment on this? Apologies if this topic has already
been discussed at length; in that case if someone could email me a brief
synopsis I would appreciate it.
--
Tom Cunningham "Good, fast, cheap -- select two."
USPS: Motorola Inc. 6501 William Cannon Dr. W. Austin, TX 78735-8598
UUCP: {ihnp4,seismo,ctvax,gatech}!ut-sally!oakhill!tomc
sun!oakhill!tomc
Phone: 512-440-2953guy%gorodish@Sun.COM (Guy Harris) (07/16/87)
> I noticed that in the latest release of Sun 3 software which we have (I > believe it is 3.2), there is no -p option for lint to perform portability > checks. This option does exist, however, in the System V version of lint > presumably distributed with the Sun 3.2 software. The "-p" option is, in fact, supplied with the System V version of "lint", but not with the 4.2 version. > Did Berkeley lint (if there is such a thing) ever have this option, was > it excised for some reason, or what? Yes, there is a version of "lint" that comes with 4BSD; in fact, there was a version of "lint" that came with V7. Both those versons have the option. > I am just curious, as I consider the constraints of eight character > symbol names (six characters external) as somewhat of an anachronism > portability-wise, but I could well be wrong. I suspect this may be why the "-p" was excised at Sun (I don't know for sure, it happened before I joined). The trouble with "lint -p" is that it's *too* strict for most uses. It checks for compatibility with some old IBM 370 and GCOS dialects of C; this is why the the 6-character *one-case* external symbol name restriction is imposed. Furthermore, instead of checking against the "lint" library for the standard C library, it checks against some small library that presumably represents the intersection of the libraries available under the C implementations at Bell Labs. (The version that comes with the S5 "lint" seems larger than the version that comes with the V7 and 4BSD "lint"s. I don't know whether this reflects additions to the OS or GCOS versions of the C library subsequent to the release of V7, or not.) The trouble is that there are some checks, performed only by "lint -p", that are useful when porting to other UNIX implementations, but specifying "-p" performs stricter checks than are useful when testing for portability to UNIX C implementations (or even to other modern C implementations). There really needs to be an option intermediate in stringency between "lint" and "lint -p"; "lint -p" itself is of limited use. Guy Harris {ihnp4, decvax, seismo, decwrl, ...}!sun!guy guy@sun.com
gwyn@brl-smoke.ARPA (Doug Gwyn ) (07/16/87)
In article <23575@sun.uucp> guy%gorodish@Sun.COM (Guy Harris) writes: >"lint -p" itself is of limited use. I agree that the UNIX System V "lint -p" is not very useful to me either. However, it would perhaps be useful in the near future for "lint" (no -p) to check for inter-POSIX portability and for "lint -p" to check for inter- X3.159 (ANSI C) portability, since those will be the two major vendor- independent portable environments of interest.