Re: Script to generate ChangeLogs automatically

[Siddhesh Poyarekar]

On 01/12/18 4:36 AM, Alfred M. Szmidt wrote:
> Tried playing with this just a bit, and please don't get the wrong
> idea with all the "it doesn't work complaints".  I think this is an
> amazing attempt at automating ChangeLogs.

Thanks, I'm looking for "it doesn't work" kind of feedback because that 
allows me to improve the script.

> But I'm a bit confundled, I tried this on one change in the glibc tree
> (see below).  While it is true that _itoa.h was included, it isn't
> really a function so it shouldn't be listed as such.

This one's pretty easy to fix.  I suppose I was just being lazy because 
the script already knows that it's a header so it only needs to use that 
information to write a more sensible output.

> The "Modified." comment seems to be redundant, since <_itoa.h> is
> included it is by definition that elf/dl-exception.c was modified so
> this should at least be "elf/dl-exception.c (_itoa.h): New.".

The "Modified" comment is for the source file as a whole.  I'll try to 
make this a bit more readable.

> I also tried running it on two recent commit on the Emacs tree, that
> got it confunlded quite badly; and produced ~740 lines of output.
> With some really weird results...  Maybe I ran it incorrectly?

It's more likely that the script is barfing; I'll check this out.

> For the format, would it make sense to maybe change it to put the hash
> in the header instead of a "COMMIT:" "thing"?  Maybe one could extend
> the format a bit to suggest this if it makes good sense.

I'm open to suggestions on the format.

I suppose I also need to push the patch up to atleast a private branch 
in glibc so that people can check out and play with the latest version 
with fixes.  I'll do that tonight.

Siddhesh