Why did we bother with line numbers at all? [closed]

Answer 1

The "Who?" would be the inventors, Kemeney and Kurtz.

After reading the replies, I checked the Wikipedia entry for "Dartmouth BASIC", and was surprised to learn

The first compiler was produced before the time-sharing system was ready. Known as CardBASIC, it was intended for the standard card-reader based batch processing system.

So, it looks like Paul Tomblin "gets the square".

Answer 2

It has a loong-loong history.

Line numbering actually comes from Dartmouth BASIC, which was the original version of the BASIC programming language and was the integral part of a so called Dartmouth Time Sharing System. That DTSS had a rudimentary IDE, which was nothing more than an interactive command line.

So every line typed inside this "IDE", and beginning with a line number, was added to the program, replacing any previously stored line with the same number; anything else was assumed to be a DTSS command and immediately executed.

Answer 3

Paul Tomblin's answer is the most comprehensive, but I'm surprised no one has mentioned that a big part of the BASIC project's initial goal was to provide a beginner-friendly interactive environment using timesharing. (Kurtz and Kemeny's vision for "universal access for all students" was far ahead of its time in this regard.)

The BASIC system that was developed to fulfill this goal featured Teletype ASR-33 (and later other) printing terminals. When connected to a timesharing-capable OS, these allowed editing and running BASIC programs in an interactive mode (unlike working with punched cards), but they are not cursor-addressable. Line numbers were a beginner-friendly way to both specify the order of program statements and allow unambiguous editing in the absence of a screen editor. The Wikipedia entry for "line editor" explains further, and anyone who's ever tried to use a line editor (such as the Un*x 'ed') can appreciate why Kurtz and Kemeny should be thanked for sparing the beginner having to learn the cryptic command sequences required for editing text in this manner.

Answer 4

In the early days, most programs were entered with punch cards. The punch cards were usually entered in sequence, usually one instruction per card, with labels (JMP/JSR targets) being a separate instruction card.

To edit your program, you replaced the card.

Later implementations added an optional sequence number on the right end of the line, so that when/if they got out of order, they could be resequenced by an automated reader.

Fortran used both numeric target labels on the left (col 1-5) and left a reserved block on the right (73-80) for sequence or comment.

When BASIC was initially written, it was decided to move the sequence numbers to the left, into FORTRAN's label field, and to allow overwriting prior cards' memory footprint... as an editing mode. This was intended for the interactive dev environment, but worked just as well with cards. And cards were used in some early implementations for a variety of reasons.

Keep in mind: Many computers were card-reader and printer interface right through the late 1970's. Even tho' interactive mode basics were available, card punched basic programs were frequently used. Since many simply were feeding into the IDE, they worked exactly the same way. Including needing a "Run" card at the end. In such cases, one could simply tack a correction card and another Run card to rerun with a variation on some variable; likewise, in complex programs, simply adding a corrected line of a card before the run was adequate to edit out problems without spending precious time finding the errant card itself.