automake
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: automake 1.4g: About `make install-strip'


From: Maciej W. Rozycki
Subject: Re: automake 1.4g: About `make install-strip'
Date: Mon, 28 May 2001 18:51:33 +0200 (MET DST)

On 26 May 2001, Tom Tromey wrote:

> Here in automake-land, we've long considered this the primary
> approach, with install-strip secondary.  My recollection is that
> install-strip was added to the standards by RMS because he didn't want
> to add INSTALL_SCRIPT.  In those days Fran├žois advocated
> INSTALL_SCRIPT precisely so that stripping could be done at install
> time.  RMS didn't like this and added install-strip instead.  For
> automake I did both: install-strip because it was mandated, and
> INSTALL_SCRIPT because, unlike install-strip, it had already been in
> use in the field for some time and was known to be useful and
> workable.

 How would install-strip solve missing INSTALL_SCRIPT?  `Install -s' might
not work on scripts and with fileutils 4.1 it does not anymore, indeed.  I
suppose there are non-GNU install implementations that fail as well.

 Actually autoconf 2.13 as released has INSTALL_SCRIPT broken (it sets it
to INSTALL_PROGRAM instead of INSTALL) but 2.50 is fine.  IMHO, the
cleanest approach for default values would actually be: 

INSTALL=<install>
INSTALL_DATA=$INSTALL -m 644
INSTALL_EXECUTABLE=$INSTALL
INSTALL_SCRIPT=$INSTALL_EXECUTABLE
INSTALL_PROGRAM=$INSTALL_EXECUTABLE

so that general executable installation options (such as permissions or an
owner) could be set in one place for both programs and scripts and
specific script or program options (such as "-s") could be set
independently.  This is an issue to be raised at the autoconf list and not
here, though. 

> I agree that having install-strip work efficiently is important.
> However I don't believe that it is more important than (1) getting 1.5
> out in a timely way (though I don't believe this will derail it either
> way), or (2) autoconf 2.13 support.

 That could be a post-1.5 item for sure.  What is now does actually work
and thus improving it is non-critical and interested parties (me included) 
might live with private patches, anyway.

> Maciej> Note that autoconf 2.13 isn't an issue anymore, I think.
> Maciej> Version 2.50 is already released.
> 
> I disagree.  I believe that uptake of 2.50 will be slow and so
> continuing support for 2.13 is important.

 OK, it looked to me like automake/autoconf 1.4/2.13 and 1.5/2.50 are
intended pairs, especially as autoconf 2.13 is really old now.  I could
have been wrong, though.

  Maciej

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: address@hidden, PGP key available        +




reply via email to

[Prev in Thread] Current Thread [Next in Thread]