gordon@gypsy.UUCP (10/15/85)
HELP !!!!!!!!!!!!! I ran across very strange errors while working with tbl and troff. My files consisted of a number of tables mixed with text. I ran each table separately through tbl and that work out all right. However, when I combined several of this tables together I got the following error from tbl: File x: line y: bad table specification character. Well, there was no error on line y, and, besides, the table was OK when I was running it through tbl separately. So, I have tried to monkey around with the file and I've noticed that if I insert a couple of blank lines between the tables, the tbl will stop complain (sometimes). Other times, deletion of a row of information in one table could stop tbl complaining about some other table. But even if the file runs through tbl some of tables get garbled when I run it through troff. Does anybody know what could possibly cause this problem ? p.s we are running Unix 4.2 on Vax 780. ------------ Igor V. Gordon Siemens Research and Technology Laboratories Princeton, NJ USENET: {ihnp4|princeton|adrvax}!siemens!gypsy!gordon ARPA: siemens!gypsy!gordon@TOPAZ
Anw%maths.nottingham.ac.uk@UCL-CS.ARPA (10/23/85)
> My files consisted of a number of tables mixed with text. I ran > each table separately through tbl and that work out all right. > However, when I combined several of this tables together I got the > following error from tbl: [...] A colleague here had a very-hard-to-find bug that sounds vaguely similar. His tables without the text ran through "tbl" and "nroff" with no problems; the text without the tables ran; the first half of the file was OK; the second half was OK; only when everything was put together did garbled output appear. This made it difficult to isolate the bug. However, I eventually spotted a ".ds ed something-or-other" just above a table; this in itself is OK, but alarm bells rang, and sure enough there was a ".ds ec ..." further up, which over-writes the built-in ".ec" request which is output by "tbl". You may have done something similar. We have now modded our "nroff"/"troff" to complain about attempts to change built-ins, and picked up several bugs in macro packages as a consequence. I suppose it is too late to convince the world that it should be an error to use or remove an undefined macro or string, or to define one that already exists; or that "," rather than "'" should be the default second control character. -- Andy Walker, Maths Dept, Nottingham Univ anw@UK.AC.Nott.Maths