Up to now I've only used glTexImage2D() with GL_RGBA for internalFormat
and format parameters. I'm trying to branch out so I looked at some
existing code. The bumps.c demo caught my eye because it uses
GL_LUMINANCE. Here's the code snippet:

bumptex = read_texture(":data:water1.bw", &texture_width,
&texture_height, &texcomps);

glTexImage2D(GL_TEXTURE_2D, 0, GL_LUMINANCE,
texture_width, texture_height, 0, GL_RGBA,
GL_UNSIGNED_BYTE, bumptex);

What puzzles me is that the format of the texture image data is
GL_RGBA, but the texture is a 256 grey (8 bit) data file. I checked
this with Graphic Converter.

Maybe I don't understand the format parameter, but doesn't the
GL_RGBA mean that the bumptex data should be in the form
R, G, B, A, R, G, B, A.... whereas we are giving it Grey1, Grey2, Grey3....

Thanks

First, read the manpage. The prototype is:

Note that there is "internal format" and "format, type". GL will convert from "format, type" to "internal format" as required. You can give it RGBA data and tell it to use it as luminance if you want; in that case it will use the R channel and ignore GBA (again, see the manpage.)

Second, you are correct, if the format is RGBA and the actual data is only 8bpp, this is an error. GL will read past the end of the buffer during upload and probably crash. However, it is possible, depending on the image loader, for the 8 bit image data to be decompressed to 32 bit RGBA as it is read from disk. You'd have to look at "bumptex" after it is loaded before calling glTexImage2D, to see how that data compares to what you see in Graphic Converter.