A short announcement: Some of you have might have already noticed the use of URLs for the Common Lisp Directory that indicate that Python code is used. It's indeed the case that we have switched from a server application based on Common Lisp to one based on Python. The reason is that we had some problems with regard to multithreading and the HTML code generation utilities that we use. In order not to risk any stability issues, we have quickly rewritten the whole application logic in Python, and it seems that everything works well now. We will give a fully detailed report of the problems of using Common Lisp in web application development in a few days when the transition is finished and fully tested.
Some of you might now think that this is a bad case of not eating our own dog food, but it's simply better to be pragmatic here and use a solution that actually works instead of spending time on fixing complex interactions between incompatible libraries. Needless to say that we stand fully committed behind Common Lisp and especially the Common Lisp Directory, due to its popularity with users and its recognition in major search engines (Google, etc.) in a very short amount of time. For the time being, we will stick to Python for the Common Lisp Directory, but will reevaluate the situation in the coming months.
Marc Battyani wrote: > It's indeed the case that we have switched from a > server application based on Common Lisp to one based on Python. The > reason is that we had some problems with regard to multithreading and > the HTML code generation utilities that we use.
So don't forget to add some notes to the Lisp frameworks you've used to the directory descriptions that they are currently crap and everybody should use Python instead ;-)
"Marc Battyani" <Marc.Batty...@fractalconcept.com> writes: > code is used. It's indeed the case that we have switched from a > server application based on Common Lisp to one based on Python. The [...] > Some of you might now think that this is a bad case of not eating our > own dog food, but it's simply better to be pragmatic here and use a > solution that actually works instead of spending time on fixing > complex interactions between incompatible libraries. Needless to say
This is indeed an unfortunate but necessary decision.
"Marc Battyani" <Marc.Batty...@fractalconcept.com> writes: > A short announcement: Some of you have might have already noticed the > use of URLs for the Common Lisp Directory that indicate that Python > code is used. It's indeed the case that we have switched from a > server application based on Common Lisp to one based on Python.
In related news, Franz announced today that they are moving their focus from Lisp to Java, due to customer demand.
Marc Battyani wrote: > A short announcement: Some of you have might have already noticed the > use of URLs for the Common Lisp Directory that indicate that Python > code is used. It's indeed the case that we have switched from a > server application based on Common Lisp to one based on Python. The > reason is that we had some problems with regard to multithreading and > the HTML code generation utilities that we use.
I was kinda hoping to do some Web app work using a CL system. Could you share a little more about the stack you abandoned, any alternative CL frameworks you rejected for what reason? [Oops. See below] I do not mean a treatise, just "we were using X (too Z), considered Y (same Z), went with (which one based on Python?)".
Also, not Ruby on Rails? The Reddit folks seemed to be close to a strong Python person (possibly even a Web app developer?) and Python was a second langugae for them so I can see them taking the quick way out.
Any plans to work your way back to CL with a project to cure the ills?
> In order not to risk > any stability issues, we have quickly rewritten the whole application > logic in Python, and it seems that everything works well now. We will > give a fully detailed report of the problems of using Common Lisp in > web application development in a few days when the transition is > finished and fully tested.
Oh. Never mind. :)
One scary thought is that I cannot imagine the site having that much traffic, so the existing tools (or at least the ones tried) must have serious problems.
"Have you ever been in a relationship?" Attorney for Mary Winkler, confessed killer of her minister husband, when asked if the couple had marital problems.
Ken Tilton <kentil...@gmail.com> wrote: > I was kinda hoping to do some Web app work using a CL system. Could you > share a little more about the stack you abandoned, [...]
I think you just got a fish sticked to your back :-)
Ken Tilton wrote: > One scary thought is that I cannot imagine the site having that much > traffic, so the existing tools (or at least the ones tried) must have > serious problems.
One of the authors of BKNR mailed me, that they have used BKNR for 2 years for a website with about 20,000 hits per hour with aserve on FreeBSD behind a squid proxy, so looks like it is possible to use Lisp for webapplications, if you carfully choose the right frameworks and program combinations. I've tried aserve, too, but probably it is not a good idea to expose it without Apache or some other proxy or redirector to the internet, because in my case aserve stopped delivering webpages after some hours.
In article <h74cq24khskl$.1ip08jo1nruti$....@40tude.net>, Frank Buss <f...@frank-buss.de> wrote:
> Ken Tilton wrote:
> > One scary thought is that I cannot imagine the site having that much > > traffic, so the existing tools (or at least the ones tried) must have > > serious problems.
> One of the authors of BKNR mailed me, that they have used BKNR for 2 years > for a website with about 20,000 hits per hour with aserve on FreeBSD behind > a squid proxy, so looks like it is possible to use Lisp for > webapplications, if you carfully choose the right frameworks and program > combinations. I've tried aserve, too, but probably it is not a good idea to > expose it without Apache or some other proxy or redirector to the internet, > because in my case aserve stopped delivering webpages after some hours.
President Clinton had the White House publications server written in Lisp running for some time on the Internet, serving american citizens with informations about his affairs. President Bush pulled the plug of that server after he got elected very fast. Go figure!
Ken Tilton <kentil...@gmail.com> writes: > I was kinda hoping to do some Web app work using a CL system. Could > you share a little more about the stack you abandoned, any alternative > CL frameworks you rejected for what reason? [Oops. See below] I do not
Marc can provide you with the technical details, but this is the tool that he finally ended up using:
So you were using a HTML generation library that was not MT safe. Since the problem was not fixed by the placement of a few good mutexes I have to assume that the library is non-reentrant. That must be embarrassing.
Wade Humeniuk wrote: > So you were using a HTML generation library that was not > MT safe. Since the problem was not fixed by the placement > of a few good mutexes I have to assume that the library is > non-reentrant. That must be embarrassing.
Man, the Reddit guys sure are off the hook. :)
Speaking of which, has this catastrophe been picked up there or on any of the other yak-yak sites?
I predict that we can add to the Lisp rap sheet (slow, interpreted, weird syntax) a new unshakeable reputation: can't even do a Web app!
Hey, Peter. How is the Lisp on Rails thing coming? :)
"Have you ever been in a relationship?" Attorney for Mary Winkler, confessed killer of her minister husband, when asked if the couple had marital problems.
"Have you ever been in a relationship?" Attorney for Mary Winkler, confessed killer of her minister husband, when asked if the couple had marital problems.
Stefan Ram wrote: > Rainer Joswig <jos...@lisp.de> writes: > >President Clinton had the White House publications server written > >in Lisp running for some time on the Internet, serving american > >citizens with informations about his affairs. President Bush pulled > >the plug of that server after he got elected very fast. Go figure!
> Is it known, whether anyone of them both is personally aware > of this, or has at least some knowledge what ťLispŤ means?
Can't say much about Bush, but atleast in Clinton's administration a small group knew, especially since they were hosting two Virtual Lisp Machines (on DEC Alpha) from 1994 to 2000 for that purpose. ;-) There were also one or two other projects done (one for the vice president Al Gore) with this technology.
>> A short announcement: Some of you have might have already noticed the >> use of URLs for the Common Lisp Directory that indicate that Python >> code is used. It's indeed the case that we have switched from a >> server application based on Common Lisp to one based on Python.
> In related news, Franz announced today that they are moving their focus > from Lisp to Java, due to customer demand.
LOL. I had a good laugh over this one. Interestingly, though, if you search on our site for the word "java", you'll get 10 pages of hits...
I suppose if you had mentioned a shift in focus from concentrating on Lisp only to concentrating on _communication_ with external langauages as well, I would have taken this much more seriously, but would have also pointed out that that was old news; we've concentrated heavily on communication with several extrnal languages for many, many years.
Friedrich Dominicus wrote: > congrats, a good one ;-)
Especially on april fool's day. For a short moment I had thought the Lispers went mad. Python for Lisp hopefully only in one particular day of the year.
Förster vom Silberwald <chain_l...@hotmail.com> wrote:
> Friedrich Dominicus wrote: >> congrats, a good one ;-)
> Especially on april fool's day. For a short moment I had thought the > Lispers went mad. Python for Lisp hopefully only in one particular day > of the year.
For me it's every day.
$ lisp ; Loading #P"/home/stesch/.cmucl-init.lisp". CMU Common Lisp 19c (19C), running on parsec With core: /usr/local/lib/cmucl/lib/lisp.core Dumped on: Fri, 2006-03-31 21:45:56+02:00 on parsec See <http://www.cons.org/cmucl/> for support information. Loaded subsystems: Python 1.1, target Intel x86 CLOS based on Gerd's PCL 2004/04/14 03:32:47
"Förster vom Silberwald" <chain_l...@hotmail.com> writes:
> Especially on april fool's day. For a short moment I had thought the > Lispers went mad. Python for Lisp hopefully only in one particular day > of the year.
>A short announcement: Some of you have might have already noticed the > use of URLs for the Common Lisp Directory that indicate that Python > code is used. It's indeed the case that we have switched from a > server application based on Common Lisp to one based on Python. The > reason is that we had some problems with regard to multithreading and > the HTML code generation utilities that we use. In order not to risk > any stability issues, we have quickly rewritten the whole application > logic in Python, and it seems that everything works well now. We will > give a fully detailed report of the problems of using Common Lisp in > web application development in a few days when the transition is > finished and fully tested.
> Some of you might now think that this is a bad case of not eating our > own dog food, but it's simply better to be pragmatic here and use a > solution that actually works instead of spending time on fixing > complex interactions between incompatible libraries. Needless to say > that we stand fully committed behind Common Lisp and especially the > Common Lisp Directory, due to its popularity with users and its > recognition in major search engines (Google, etc.) in a very short > amount of time. For the time being, we will stick to Python for the > Common Lisp Directory, but will reevaluate the situation in the > coming months.
OK finally, we decided on April 2 to go back to the Lisp version ;-) So now the Common Lisp Directory is back to: http://www.cl-user.net/asp/root-dir (BTW this makes me thinks that I should drop the asp as well... :)
In fact as probably anyone has found now, the Common Lisp Directory has never been rewritten in Python. It was just an April 1 joke for insiders (remember the R affair...)
A more detailed report on the Common Lisp Directory experience will follow.
"Marc Battyani" <Marc.Batty...@fractalconcept.com> wrote
>A short announcement: Some of you have might have already noticed the > use of URLs for the Common Lisp Directory that indicate that Python > code is used.
As most of you have probably found out now, the switch of the Common Lisp Directory to Python was only effective on April First. ;-)
Here are some feedback and real life data from the Common Lisp Directory project.
Some data taken on April 2 (in decreasing order :):
42 041 200 Hits since the Dec 15 2005 (start of the linkit project)!
41 997 354 Hits served by the same Lisp process, the other ones are the logo/jpg etc. (with a max at 930K hits/day and several days with more than 800K hits)
First some explanation about why so many hits: As some of you already know, my framework is intended for complex real time collaborative applications and for that there is a 2 seconds Ajax like keep alive/bidirectional connection. In the case of a public application like the cl-directory this continuous connection is not useful but induces a lots of hits. For now, I've reduced the connection frequency to 5 seconds and I will suppress it completely for all the non interactive pages in the future.
Anyway this highlights the rock solid stability and reliability of a Lisp webapp. The Lisp image is the same one from December. It has never crashed, though the application have evolved from linkit (an improved reddit like app) to the cl-directory with several versions that have completely changed the applicative and object model.
> server application based on Common Lisp to one based on Python. The > reason is that we had some problems with regard to multithreading and > the HTML code generation utilities that we use.
On the contrary, the hability to blend the HTML generation in the Lisp code through HTML macros is far better than every alternative that I know of.
[...]
> Needless to say > that we stand fully committed behind Common Lisp and especially the > Common Lisp Directory, due to its popularity with users and its > recognition in major search engines (Google, etc.) in a very short > amount of time.
At least that part of the message is true. After only 2.5 month with some real content, the Common Lisp Directory already consistently appears in the very first links for a lot of common lisp related searches on Google. Sometimes it's even the first one!
So don't worry, the CLD will stay in Common Lisp and will continue to improve. The next version (when I find some time to play with this) will enable anybody to edit the CLD content but still in a moderated way to avoid being spammed
In the mean time you can already add notes and comments to the directory pages and of course submit new pages.
BTW some people have complained that they could not edit their own libraries/persons/etc. entries. In that case just send us an email with your cl-directory user login name and we will give you the modification rights you want.
As a conclusion, I would say that Common Lisp has once again proven to be a very good way to very quickly build rock solid and reliable web applications.
Why do you use an Apache module for forwarding requests with another socket connection from the Apache module to the Lisp server instead of accepting the internet requests in Lisp, only?
