<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 TRANSITIONAL//EN"> <HTML> <HEAD> <META HTTP-EQUIV="Content-Type" CONTENT="text/html; CHARSET=UTF-8"> <META NAME="GENERATOR" CONTENT="GtkHTML/3.32.2"> </HEAD> <BODY> On Tue, 2011-11-08 at 23:34 +0000, Matthieu Herrb wrote: <BLOCKQUOTE TYPE=CITE> <TT>On Tue, Nov 08, 2011 at 04:05:06PM -0500, Gaetan Nadon wrote:</TT> <TT>> On Mon, 2011-11-07 at 19:57 +0000, Matthieu Herrb wrote:</TT> <TT>> </TT> <TT>> > This logically reverts 7e223d3ac6c0d549a7d6e4dcc86a053e19594028.</TT> <TT>> > There are still cases (in particular OpenBSD) where the shared greeter</TT> <TT>> > is not desired.</TT> <TT>> </TT> <TT>> My reading at the time was the static version was there because some O/S</TT> <TT>> did not have loadable libraries support (this greeter seems to be very</TT> <TT>> old code). Can you add in the commit text (if you know) the reasons why</TT> <TT>> OpenBSD would not be able to use a shared library? Less chances of being</TT> <TT>> removed in a few years from now. I thought all platforms could use</TT> <TT>> shared libraries.</TT> <TT>OpenBSD/vax (don't laugh please) is still a.out and doesn't have</TT> <TT>shared libs. Yet it can run xdm (no X server though since the ability</TT> <TT>of statically linking the drivers was lost a few years ago :().</TT> </BLOCKQUOTE> Fair enough, consider adding this info in the commit text. <BLOCKQUOTE TYPE=CITE> <TT>But this is not the main reason. Basically the shared object doesn't</TT> <TT>give us any benefit. So why bother with the extra complexity and</TT> <TT>failures cases that it introduces ?</TT> </BLOCKQUOTE> I had considered doing just that at the time. I recall having seen on the net a site showing how to create a custom greeter. That made me choose shared vs static having the (wrong) assumption that all platforms supported shared libs. <BLOCKQUOTE TYPE=CITE> <TT>> </TT> <TT>> > </TT> <TT>> > BTW, Ubuntu 11.10 and Fedora 15 both ship xdm 1.10 with the shared</TT> <TT>> > greeter disabled too.</TT> <TT>> </TT> <TT>> And we don't if there are reasons for that or if it is just an old build</TT> <TT>> script that works...</TT> <TT>As Alan pointed out, there is no real reason to keep the shared object</TT> <TT>version as long as no one provides alternate greeters that could be</TT> <TT>plugged in to replace the default one.</TT> <TT>If you want to shrink the source, I would rather drop the support for</TT> <TT>the dynamic greeter.</TT> </BLOCKQUOTE> I am not opposed at all to a static only build. Perhaps a next patch should do that. This would certainly reduce module complexity. <BLOCKQUOTE TYPE=CITE> </BLOCKQUOTE> </BODY> </HTML>