You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Sander Striker <st...@apache.org> on 2001/12/07 18:39:15 UTC
RE: cvs commit: httpd-2.0 ROADMAP
> From: Ryan Bloom [mailto:rbb@covalent.net]
> Sent: 07 December 2001 17:22
> To: dev@httpd.apache.org; brianp@apache.org; httpd-2.0-cvs@apache.org
> Subject: Re: cvs commit: httpd-2.0 ROADMAP
>
>
>
> > + * Add a string "class" that combines a char* with a length
> > + and a reference count. This will help reduce the number
> > + of strlen and strdup operations during request processing.
>
> This doesn't belong in Apache, if anything it is an APR class. BTW,
> this has come up multiple times, and everybody seems to be in favor
> or if.
>
> Ryan
In subversion svn_string_t was introduced, because it wasn't in APR.
It does (AFAIK) exactly what brian describes.
Sander
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: cvs commit: httpd-2.0 ROADMAP
Posted by "William A. Rowe, Jr." <wr...@covalent.net>.
From: "Brian Pane" <br...@cnet.com>
Sent: Friday, December 07, 2001 12:54 PM
> Ian Holsman wrote:
>
> >On Fri, 2001-12-07 at 10:39, Sander Striker wrote:
> >
> ...
>
> >>In subversion svn_string_t was introduced, because it wasn't in APR.
> >>It does (AFAIK) exactly what brian describes.
> >>
> >is there any reason NOT to introduce this into the APR now?
> >we could then slowly migrate strings to this
> >
>
> I'm in favor of including it in APR, but opposed to trying to
> retrofit it into Apache before 2.0 GA. My rationale for this
> is that, even with a reference-counted string class, we'll have
> to be extremely careful not to make performance worse when doing
> a major change in the string handling. (And we're close to having
> all the really significant strlen and strdup inefficiencies fixed
> in httpd-2.0 already.)
+1 ... a STRINGS branch in apr would be good while we experiment.
Re: cvs commit: httpd-2.0 ROADMAP
Posted by "William A. Rowe, Jr." <wr...@covalent.net>.
From: "Brian Pane" <br...@cnet.com>
Sent: Friday, December 07, 2001 12:54 PM
> Ian Holsman wrote:
>
> >On Fri, 2001-12-07 at 10:39, Sander Striker wrote:
> >
> ...
>
> >>In subversion svn_string_t was introduced, because it wasn't in APR.
> >>It does (AFAIK) exactly what brian describes.
> >>
> >is there any reason NOT to introduce this into the APR now?
> >we could then slowly migrate strings to this
> >
>
> I'm in favor of including it in APR, but opposed to trying to
> retrofit it into Apache before 2.0 GA. My rationale for this
> is that, even with a reference-counted string class, we'll have
> to be extremely careful not to make performance worse when doing
> a major change in the string handling. (And we're close to having
> all the really significant strlen and strdup inefficiencies fixed
> in httpd-2.0 already.)
+1 ... a STRINGS branch in apr would be good while we experiment.
Re: cvs commit: httpd-2.0 ROADMAP
Posted by Brian Pane <br...@cnet.com>.
Ian Holsman wrote:
>On Fri, 2001-12-07 at 10:39, Sander Striker wrote:
>
...
>>In subversion svn_string_t was introduced, because it wasn't in APR.
>>It does (AFAIK) exactly what brian describes.
>>
>is there any reason NOT to introduce this into the APR now?
>we could then slowly migrate strings to this
>
I'm in favor of including it in APR, but opposed to trying to
retrofit it into Apache before 2.0 GA. My rationale for this
is that, even with a reference-counted string class, we'll have
to be extremely careful not to make performance worse when doing
a major change in the string handling. (And we're close to having
all the really significant strlen and strdup inefficiencies fixed
in httpd-2.0 already.)
--Brian
Re: cvs commit: httpd-2.0 ROADMAP
Posted by Brian Pane <br...@cnet.com>.
Ian Holsman wrote:
>On Fri, 2001-12-07 at 10:39, Sander Striker wrote:
>
...
>>In subversion svn_string_t was introduced, because it wasn't in APR.
>>It does (AFAIK) exactly what brian describes.
>>
>is there any reason NOT to introduce this into the APR now?
>we could then slowly migrate strings to this
>
I'm in favor of including it in APR, but opposed to trying to
retrofit it into Apache before 2.0 GA. My rationale for this
is that, even with a reference-counted string class, we'll have
to be extremely careful not to make performance worse when doing
a major change in the string handling. (And we're close to having
all the really significant strlen and strdup inefficiencies fixed
in httpd-2.0 already.)
--Brian
Re: cvs commit: httpd-2.0 ROADMAP
Posted by "William A. Rowe, Jr." <wr...@covalent.net>.
From: "Aaron Bannert" <aa...@clove.org>
Sent: Friday, December 07, 2001 12:46 PM
> [narrowing down the to: list back to APR]
>
> On Fri, Dec 07, 2001 at 10:44:14AM -0800, Ian Holsman wrote:
> > On Fri, 2001-12-07 at 10:39, Sander Striker wrote:
> > > In subversion svn_string_t was introduced, because it wasn't in APR.
> > > It does (AFAIK) exactly what brian describes.
> > >
> > is there any reason NOT to introduce this into the APR now?
> > we could then slowly migrate strings to this
>
> +1
> I see no reason not to start adding this, as long as it's marked
> in some way to be experimental/incomplete while we're working on it.
'marked in some way' == branch.
> I assume we'll have to start duping some of our string functions to
> be able to use the new type?
But of course ... and provide the a complete set of ISO C string functions?
Re: cvs commit: httpd-2.0 ROADMAP
Posted by Aaron Bannert <aa...@clove.org>.
[narrowing down the to: list back to APR]
On Fri, Dec 07, 2001 at 10:44:14AM -0800, Ian Holsman wrote:
> On Fri, 2001-12-07 at 10:39, Sander Striker wrote:
> > In subversion svn_string_t was introduced, because it wasn't in APR.
> > It does (AFAIK) exactly what brian describes.
> >
> is there any reason NOT to introduce this into the APR now?
> we could then slowly migrate strings to this
+1
I see no reason not to start adding this, as long as it's marked
in some way to be experimental/incomplete while we're working on it.
I assume we'll have to start duping some of our string functions to
be able to use the new type?
-aaron
RE: cvs commit: httpd-2.0 ROADMAP
Posted by Ian Holsman <ia...@apache.org>.
On Fri, 2001-12-07 at 10:39, Sander Striker wrote:
> > From: Ryan Bloom [mailto:rbb@covalent.net]
> > Sent: 07 December 2001 17:22
> > To: dev@httpd.apache.org; brianp@apache.org; httpd-2.0-cvs@apache.org
> > Subject: Re: cvs commit: httpd-2.0 ROADMAP
> >
> >
> >
> > > + * Add a string "class" that combines a char* with a length
> > > + and a reference count. This will help reduce the number
> > > + of strlen and strdup operations during request processing.
> >
> > This doesn't belong in Apache, if anything it is an APR class. BTW,
> > this has come up multiple times, and everybody seems to be in favor
> > or if.
> >
> > Ryan
>
> In subversion svn_string_t was introduced, because it wasn't in APR.
> It does (AFAIK) exactly what brian describes.
>
is there any reason NOT to introduce this into the APR now?
we could then slowly migrate strings to this
> Sander
RE: cvs commit: httpd-2.0 ROADMAP
Posted by Sander Striker <st...@apache.org>.
> From: Karl Fogel [mailto:kfogel@newton.ch.collab.net]
> Sent: 07 December 2001 19:42
> "Sander Striker" <st...@apache.org> writes:
> > > > + * Add a string "class" that combines a char* with a length
> > > > + and a reference count. This will help reduce the number
> > > > + of strlen and strdup operations during request processing.
> > >
> > > This doesn't belong in Apache, if anything it is an APR class. BTW,
> > > this has come up multiple times, and everybody seems to be in favor
> > > or if.
> > >
> > > Ryan
> >
> > In subversion svn_string_t was introduced, because it wasn't in APR.
> > It does (AFAIK) exactly what brian describes.
>
> Subversion's `svn_string_t' and `svn_stringbuf_t' don't have reference
> counts...
>
> -Karl
True, and having a refcount would mean they would need to come out of some
global 'pool'. The pool passed in for creation/duplication of the string
merely bumps the refcount and registers a 'string refcount decrease [delete
when zero]' cleanup. At least, that is what pops in my mind at first thought.
Refcounts and pools are hairy... :)
Caching the strlen is a good start though.
Sander
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
RE: cvs commit: httpd-2.0 ROADMAP
Posted by Sander Striker <st...@apache.org>.
> From: Karl Fogel [mailto:kfogel@newton.ch.collab.net]
> Sent: 07 December 2001 19:42
> "Sander Striker" <st...@apache.org> writes:
> > > > + * Add a string "class" that combines a char* with a length
> > > > + and a reference count. This will help reduce the number
> > > > + of strlen and strdup operations during request processing.
> > >
> > > This doesn't belong in Apache, if anything it is an APR class. BTW,
> > > this has come up multiple times, and everybody seems to be in favor
> > > or if.
> > >
> > > Ryan
> >
> > In subversion svn_string_t was introduced, because it wasn't in APR.
> > It does (AFAIK) exactly what brian describes.
>
> Subversion's `svn_string_t' and `svn_stringbuf_t' don't have reference
> counts...
>
> -Karl
True, and having a refcount would mean they would need to come out of some
global 'pool'. The pool passed in for creation/duplication of the string
merely bumps the refcount and registers a 'string refcount decrease [delete
when zero]' cleanup. At least, that is what pops in my mind at first thought.
Refcounts and pools are hairy... :)
Caching the strlen is a good start though.
Sander
RE: cvs commit: httpd-2.0 ROADMAP
Posted by Sander Striker <st...@apache.org>.
> From: Karl Fogel [mailto:kfogel@newton.ch.collab.net]
> Sent: 07 December 2001 19:42
> "Sander Striker" <st...@apache.org> writes:
> > > > + * Add a string "class" that combines a char* with a length
> > > > + and a reference count. This will help reduce the number
> > > > + of strlen and strdup operations during request processing.
> > >
> > > This doesn't belong in Apache, if anything it is an APR class. BTW,
> > > this has come up multiple times, and everybody seems to be in favor
> > > or if.
> > >
> > > Ryan
> >
> > In subversion svn_string_t was introduced, because it wasn't in APR.
> > It does (AFAIK) exactly what brian describes.
>
> Subversion's `svn_string_t' and `svn_stringbuf_t' don't have reference
> counts...
>
> -Karl
True, and having a refcount would mean they would need to come out of some
global 'pool'. The pool passed in for creation/duplication of the string
merely bumps the refcount and registers a 'string refcount decrease [delete
when zero]' cleanup. At least, that is what pops in my mind at first thought.
Refcounts and pools are hairy... :)
Caching the strlen is a good start though.
Sander
Re: cvs commit: httpd-2.0 ROADMAP
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
"Sander Striker" <st...@apache.org> writes:
> > > + * Add a string "class" that combines a char* with a length
> > > + and a reference count. This will help reduce the number
> > > + of strlen and strdup operations during request processing.
> >
> > This doesn't belong in Apache, if anything it is an APR class. BTW,
> > this has come up multiple times, and everybody seems to be in favor
> > or if.
> >
> > Ryan
>
> In subversion svn_string_t was introduced, because it wasn't in APR.
> It does (AFAIK) exactly what brian describes.
Subversion's `svn_string_t' and `svn_stringbuf_t' don't have reference
counts...
-Karl
Re: cvs commit: httpd-2.0 ROADMAP
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
"Sander Striker" <st...@apache.org> writes:
> > > + * Add a string "class" that combines a char* with a length
> > > + and a reference count. This will help reduce the number
> > > + of strlen and strdup operations during request processing.
> >
> > This doesn't belong in Apache, if anything it is an APR class. BTW,
> > this has come up multiple times, and everybody seems to be in favor
> > or if.
> >
> > Ryan
>
> In subversion svn_string_t was introduced, because it wasn't in APR.
> It does (AFAIK) exactly what brian describes.
Subversion's `svn_string_t' and `svn_stringbuf_t' don't have reference
counts...
-Karl
RE: cvs commit: httpd-2.0 ROADMAP
Posted by Ian Holsman <ia...@apache.org>.
On Fri, 2001-12-07 at 10:39, Sander Striker wrote:
> > From: Ryan Bloom [mailto:rbb@covalent.net]
> > Sent: 07 December 2001 17:22
> > To: dev@httpd.apache.org; brianp@apache.org; httpd-2.0-cvs@apache.org
> > Subject: Re: cvs commit: httpd-2.0 ROADMAP
> >
> >
> >
> > > + * Add a string "class" that combines a char* with a length
> > > + and a reference count. This will help reduce the number
> > > + of strlen and strdup operations during request processing.
> >
> > This doesn't belong in Apache, if anything it is an APR class. BTW,
> > this has come up multiple times, and everybody seems to be in favor
> > or if.
> >
> > Ryan
>
> In subversion svn_string_t was introduced, because it wasn't in APR.
> It does (AFAIK) exactly what brian describes.
>
is there any reason NOT to introduce this into the APR now?
we could then slowly migrate strings to this
> Sander
Re: cvs commit: httpd-2.0 ROADMAP
Posted by Karl Fogel <kf...@newton.ch.collab.net>.
"Sander Striker" <st...@apache.org> writes:
> > > + * Add a string "class" that combines a char* with a length
> > > + and a reference count. This will help reduce the number
> > > + of strlen and strdup operations during request processing.
> >
> > This doesn't belong in Apache, if anything it is an APR class. BTW,
> > this has come up multiple times, and everybody seems to be in favor
> > or if.
> >
> > Ryan
>
> In subversion svn_string_t was introduced, because it wasn't in APR.
> It does (AFAIK) exactly what brian describes.
Subversion's `svn_string_t' and `svn_stringbuf_t' don't have reference
counts...
-Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org