on egcs-1.0

on egcs-1.0

[Orn E. Hansen]

  I've read some talks about the release, egcs-1.0.  Since I'm reading 
that if I apply patches from 1201 to 1.0, it won't be compatible with
the next snapshot.  I wonder about binary compatibility of programs
compiled with one version with the other. Will the difference between
the versions affect a program, that is dynamically compiled to one, on
the other (I expect both to have a libstdc++.so.2.8.0 name).

  Since I have a very slow link, I'm not very keen on downloading
patches or upgrades, unless they are necessary or provide a needed
functionality... or solve a critical bug.  And 1201 snapshot is
working beautifully here now.

Re: on egcs-1.0

[Jeffrey A Law]

  In message < 199712071114.MAA02853@oehansen.pp.se >you write:
  > 
  >   I've read some talks about the release, egcs-1.0.  Since I'm reading 
  > that if I apply patches from 1201 to 1.0, it won't be compatible with
  > the next snapshot.  I wonder about binary compatibility of programs
  > compiled with one version with the other. Will the difference between
  > the versions affect a program, that is dynamically compiled to one, on
  > the other (I expect both to have a libstdc++.so.2.8.0 name).
I don't think we're introducing any binary incompatabilities at this
time.

At some point we will, and I hope the C++ folks will warn everyone
about it when it happens.

jeff

Re: on egcs-1.0

[Martin von Loewis]

> I don't think we're introducing any binary incompatabilities at this
> time.
> 
> At some point we will, and I hope the C++ folks will warn everyone
> about it when it happens.

The original wording using 'incompatible' was somewhat confusing.
What it (probably?) meant is: The next snapshot will include a patch
relative to 1201, not to 1.0. So if you apply the 1.0 patch now, you
will not be able to patch to the next snapshot, unless you reconstruct
the 1201 sources first.

Regards,
Martin

Re: on egcs-1.0

[Jeffrey A Law]

  In message < 199712072306.AAA18675@mira.isdn.cs.tu-berlin.de >you write:
  > The original wording using 'incompatible' was somewhat confusing.
Yup.

  > What it (probably?) meant is: The next snapshot will include a patch
  > relative to 1201, not to 1.0. So if you apply the 1.0 patch now, you
  > will not be able to patch to the next snapshot, unless you reconstruct
  > the 1201 sources first.
That is the case.  I've mentioned it several times to this list.

jeff