Re: [flashrom] [PATCH 2/4] Refactor PCI and USB device status printing.

[Carl-Daniel Hailfinger]

Am 26.12.2012 20:44 schrieb Stefan Tauner:
> On Wed, 26 Dec 2012 20:32:35 +0100
> Carl-Daniel Hailfinger  wrote:
> > Can't you simply handle USB/PCI/OTHER without switch()? After all,
> > devs.pci, devs.usb and devs.note are in a union and you can't detect
> > bugs like .type=USB combined with .devs.note="foobar" anyway. The only
> > special case is the internal programmer. Suggestion:
> >
> > switch(p.type) {
> > case USB:
> > case PCI:
> > case OTHER:
> >     /* This is a union, check only one member. */
> >     if (p.devs.note == NULL) {
> >         if (strcmp("internal", p.name) == 0)
> >             break; /* This one has its device list stored separately. */
> >         msg_gerr("Programmer %s has neither a device list nor a textual
> > description!\n", p.name);
> >         ret = 1;
> >     break;
> > default:
> > ...
>
> that's certainly possible. the rationale for my code (besides its
> development while hacking around and refining) is:
> the detailed messages could help the programmer writer better than a
> generic message that catches all.
> the question is if that is worth it...

I don't think it's worth it. My suggested error message is not optimal,
though. It may have been misunderstood as "please add a device list or
textual description, your choice" which would be clearly wrong. What about

"Programmer %s has no device list/textual description of the right type!\n"

Regards,
Carl-Daniel