From: cabr85 AT ccsun DOT strath DOT ac DOT uk ("p.a.c.tavares") Subject: bug report 17 Feb 1997 11:10:24 -0800 Approved: cygnus DOT gnu-win32 AT cygnus DOT com Distribution: cygnus Message-ID: <199702171624.QAA10445.cygnus.gnu-win32@avon.cc.strath.ac.uk> Content-Type: text Original-To: gnu-win32 AT cygnus DOT com (Cygwin List) X-Mailer: ELM [version 2.4 PL24] Original-Sender: owner-gnu-win32 AT cygnus DOT com The following code works fine in a HP system (using gcc 2.7.2), calling the correct "Column" contructor with the right arguments int Table::init(int line,int col,double fill_value) { n_lines = line; n_cols = col; if (pt) delete [] pt; pt = new Column[col](line,fill_value); if (pt == NULL) return 1; return 0; } However under cygwin, in spite of compiling fine it calls (according to gdb) the correct column constructor "Column::Column(int,double) but all the arguments show up as 0 (including the int which results in a lenght 0 column being created). The form of this contructor is I believe irrelevant but here it is anyway. A check of the value of 'size' in the first line of the function tells that size=0 (when it was called with a positive number by Table::init(int,double), Column::Column(int size,double fill_value) { len = size; val = new float [len]; while (size > 0) val[--size] = fill_value; } I am using gcc in a win95 system, 486 with 16Mb. the options of the compiler are -gstabs+ and a -I entry. The libraries are, -lstdc++ -lg++ -lm -lpact (the last being my own library) ---------------------------------------------------------------------------- Pedro A.C. Tavares

Superconducting Devices Group Tel: +44-141-548 3151 Dept. of Physics & Applied Physics Fax: +44-141-552 7143 University of Strathclyde For PGP public key: "finger Glasgow, UK pedro_t AT zebedee DOT phys DOT strath DOT ac DOT uk" ---------------------------------------------------------------------------- - For help on using this list, send a message to "gnu-win32-request AT cygnus DOT com" with one line of text: "help".