This is the mail archive of the cygwin-apps mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: setup


On Jun 10 22:50, Achim Gratz wrote:
> Corinna Vinschen writes:
> >> > Either way, please go ahead and apply your patch.  While you're at it,
> 
> Done.
> 
> >> > would you mind to change YY_READ_BUF_SIZE as above and raise the buffer
> >> > size in io_stream::copy to 64K as well?
> >>
> >> Yes, but I'll probably put each of those in a separate commit.
> >
> > Sure, thank you.
> 
> I'll do that later since I have to re-shuffle some local commits and
> then test everything again first.
> 
> 
> If you're so inclined you could perhaps have a look at the next commit
> on my local branch:
> 
> http://repo.or.cz/w/cygwin-setup/local.git/commitdiff/5b4e8b928c7fe16b2813b5f1bdbf27b49c8e0d57
> 
> This implements xz compressed setup.xz files over network in addition to
> setup.bz2.  The difference between bz2 and xz is pretty consistently
> about 10% smaller files for xz, but the main envisioned advantage some
> time down the road (years, most likely) would be to get rid of the
> additional compression libraries in setup once everything is compressed
> with xz.  The best compression setting with the setup files I have
> looked at is "-6e".  I've also checked that setup can deal correctly
> with different checksums for the xz blocks (standard is crc64, but
> sha256 works).

Ok, I took a look.  Basically it looks fine, but it seems there's a bug
in terms of sig_fail.

You're giving sig_fail as argument to fetch_remote_ini.  Inside
fetch_remote_ini, you're setting sig_fail, but you never test it.
However, since sig_fail has been given to fetch_remote_ini by value,
not by reference, the caller will never see the change to sig_fail.
So in the expression

  if (!ini_file && !sig_fail)

sig_fail is always false.

I'm also wondering of the new logic at this point.  If you fix the above
call by value of sig_fail, the new logic will only continue if there
was no sig test fail.  It will not continue if there was no ini file,
as it did before.  Why?


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

Attachment: pgpFyK14HYoMj.pgp
Description: PGP signature


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]