Re: Back on track

From: Daniel Stenberg <Daniel.Stenberg_at_sth.frontec.se_at_hypermail-project.org>
Date: Fri, 18 Dec 1998 07:44:29 +0100 (MET)
Message-ID: <Pine.SO4.4.05.9812180732190.11941-100000_at_wcsw110>

On Thu, 17 Dec 1998, Kent Landfield wrote:

> o The version that was used as the basis (hypermail.981013) was
> not anywhere near the version that I had been working on. That
> had been a version 10 days earlier. There were a good deal of
> date corrections, HTML parsing corrections, memory overwrite
> corrections, etc.

... and 2b12 is also more than 10 days work from the 981013 version.

> o There are some issues that need to be discussed a bit before
> we use dynamic strings everywhere in hypermail. There are
> places where string usage is quite constrained and well behaved
> and dynamic string usage is not a win. There are however places
> where is it drastically needed such as in message parsing and
> HTML generation. Performance and stability are the big considerations
> in all this.

There is no need for dynamic strings where we can use static limited buffers. But we have to be absolutely certain that the buffer will never ever need or want to be bigger.

I don't think you'll find any serious "abuse" of dynamic strings in my version. Besides, my use of dynamic strings is free to optimize and so are the functions I use to build dynamic strings with.

> So here is what I plan to do.... I am going to grab a few things out that
> have been added during my absense. These are low hanging fruit that can
> be quickly integrated into 2.0b4. I will then release that version.

I hope you will learn from all this, that we want very frequent releases so that others can contribute at almost any one time, starting with the most recently released version.

> It has had a good deal of work done to it and it should be the basis for
> going forward. At that point we can look at adding validated html
> parsing and dynamic string usage.

I'm dissapointed you removed all my sweat and tears that easily. What good is b4 without dynamic strings? In my view, more than 50% of all bugs we get to hear about here on the list is about buffer overflows. My version is really close to have fixed all of those problems.

I would've suggested that you put your changes on the 2b12 version since my changes were all over and change a lot of the infrastructure because of the dynamic strings.

I would've appriciated a little discussion about this before you did this.

Now what, should I re-insert all my fixes onto the b4 or do I wait until you do that?

-- 
   Daniel Stenberg    http://www.fts.frontec.se/~dast     0708-317742
   ech`echo xiun|tr nu oc|sed 'sx\([sx]\)\([xoi]\)xo un\2\1 is xg'`ol

Received on Fri 18 Dec 1998 08:48:23 AM GMT