|
|
Magic Mailing List |
|
From: R. Timothy Edwards (tim AT stravinsky DOT jhuapl.edu) Date: Mon Nov 11 2002 - 17:14:37 EST
Dear Stefan, The "flock_list" directory is a result of selecting the "file locking" option when doing the "make config" stage at the beginning of the compile and install process. If it's producing an error, it's because that particular version of the file locking is broken, although the only effect you should notice is the error message. The best solution to that one is either to 1) ignore it, 2) recompile with the "file locking" option disabled, or 3) download the most recent CVS source of magic 7.1, and compile that (either with or without file locking, but at least the implementation is correct). > scn6m.tech27: line 2385: section cifoutput: > Reducer for lambda=0.10(p) CIF doesn't divide distances evenly. This is not precisely an error; it's a minor problem with the technology file. The CIF "reducer" factor is supposed to make CIF files smaller by dividing all layout measurements by some factor (in this case, 5; the tech file, assuming it's the same as SCN6M_SUBM.10.tech27, has the line: scalefactor 10 5 in which the "10" is the scale (lambda=0.10), and "5" is the reducer. This assumes, then, that all numbers generated by CIF will be divisible by 5. The SCN6M_SUBM.10.tech27 doesn't have any such values, but your techfile might. If so, then you are being warned by magic that expected output values in CIF output may be rounded off. If you write GDS, this can be ignored. If you write CIF, you should change the reducer in the scalefactor line of the tech file to "1" to be on the safe side. I recognized the stupidity of having to make the tech file maintainer "do the math" to figure out what the greatest common denominator of all CIF grow/shrink/bloat/squares values should be, so I changed this in magic version 7.2 so that the correct reducer is computed automatically and ignored where it appears in the tech file. Your version of magic will also give an error if the scalefactor is not evenly divisibly by 2 times the reducer. This is a bogus error message (the extra factor of 2 compensates for the fact that CIF reports rectangles by their centers, so rectangles with odd-valued widths or heights may end up with centers on 0.5-unit boundaries. This factor of 2 is already added to the CIF output, and is independent of the reducer. The error message is apparently left over from a long time ago). Hopefully this answers your questions. Regards, Tim
|
|
|
|