Subject: Re: [libssh2] Where the hell has Sara been?

Re: [libssh2] Where the hell has Sara been?

From: Daniel Stenberg <daniel_at_haxx.se>
Date: Sun, 12 Nov 2006 11:36:40 +0100 (CET)

On Sat, 11 Nov 2006, Sara Golemon wrote:

> Hi everyone. That was a busy week there. Y'all would just wait till the
> ONE week I'm out of the country to start makin' noise, wouldn't ya?

Nah, I've been lurking here waiting for you to leave before I jumped out and
spammed this list! ;-P

> * Yes, those usually are religious issues.
> * I don't see anything wrong with placing hard-wraps at 80 chars and it
> probably would add much to readability in most cases. Let's make it a soft
> rule though, there are times when a couple extra over is better than a forced
> split.
> * The only rules I'm really particular about are:
> * Tab-indents, not spaces.
> * Bracing, even for one-line if conditions
> * Commenting commenting commenting :)
> * Yes, I know I've broken all three of these rules at least once :)

I would certainly be fine with these rules, and them being "soft" feels like
the only sensible approach. We could put them into a HACKING document to make
it easy to figure out for newcomers (such as myself).

> Moving private include files

> * Is there any particular reason this is desired? I'm not inherently
> against it, but I don't see an impetus for it.

The main reasons would be:

1 - private header files don't need the libssh2 name prefix

2 - it makes it a lot easier to understand which headers that are public and
     thus installed with 'make install' and which that aren't

3 - I find it more logical to keep private source-related stuff in the source
     code dir. Other projects seem to adhere to this layout as well.

> 32bit variables:
> * Yeah, I know treating long as 32bit is an abuse, but I think I've taken
> reasonable steps to ensure that overflows aren't problematic.

Yeah, I couldn't spot any obvious problem in the code but I think using longs
like that make it easier that an error slips in when the code is edited. IMO.

> * In theory someone might try building on a 16bit device (some embedded
> app), and this is reason enough to properly detect a true 32bit type, but I
> won't lose much sleep over it.

I don't think its worth persuing any goals to remain functional on 16bit
machines, and they are hardly existing anymore. Embedded devices are mosly
32bit as well these days, at least when they would be considering using
libssh2 in the first place.

> * The win32 build system is something someone else threw in as I have
> neither the experience or the motivation to focus on DLL builds. If someone
> wants to make this area shine, you'd be a godsend.

I hope to be able to work on win32 builds cross-compiled on Linux soon. It
isn't high prio for me either though.

> * Sure, go for it. In the beginning a hand-crafted Makefile.in was simpler
> to get things going, but there are a few source files and the file is
> getting a bit unruly. Using a proper automake setup should reduce some of
> those problems.

Yeah, the Makfiles do get bigger and more hard to read etc, but we get the
libtool stuff in there and some other neat stuff, basically automaticly, so I
think that speaks in favour of taking this step.

> CVS Acccess:
> * Who wants it?

For the record, I've applied.

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
libssh2-devel mailing list
libssh2-devel_at_lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/libssh2-devel
Received on 2006-11-12