Lisp for web apps? Tell me more.. (seriously)
30 comments
Here's a detailed link on setting up a scheme server (perhaps more detailed than many system admins would need). I'd err towards mod_lisp under Apache rather than run SISC Scheme under Tomcat, but this is a pleasant read:
http://www.lisperati.com/quick.html
Why did you decide to go with Scheme as opposed to Lisp? (I'm very curious about the pros and cons)
Scheme is a dialect of Lisp.
I know - what benefits do you get from Scheme over CL?
I can't speak for the GP, but since it has been 11 hours and he haven't answered yet I'll give you my answer: it's mostly a matter of taste, or sometimes if there is a particular library/app/framework for one which you'd like to use.
I would also add that at least in my experience the Scheme community was much friendlier the the Common Lisp one. They don't particularly like novices over there and it is quite easy to get flamed if you're not careful enough...
There are multiple CL communities. comp.lang.lisp famously sucks, but (for instance) #lisp isn't so bad.
Here's an overview of different approaches: http://bc.tech.coop/blog/041017.html
And here's a list of different web server resources: http://www.cl-user.net/asp/tags/web-servers
And here's a list of different web server resources: http://www.cl-user.net/asp/tags/web-servers
The libraries we're currently using:
Portable AllegroServe: Lisp Webserver (with Webactions Session control) http://sourceforge.net/projects/portableaserve/
ASDF: Another System Definition Facility http://www.cliki.net/asdf
CLSQL: Database interface Library http://www.cliki.net/CLSQL
UFFI: Universal Foreign Function Interface http://www.cliki.net/UFFI
Portable AllegroServe: Lisp Webserver (with Webactions Session control) http://sourceforge.net/projects/portableaserve/
ASDF: Another System Definition Facility http://www.cliki.net/asdf
CLSQL: Database interface Library http://www.cliki.net/CLSQL
UFFI: Universal Foreign Function Interface http://www.cliki.net/UFFI
About 60 Web clients (web client), servers, application servers, browsers, etc:
http://www.cliki.net/web
There's also this resource which is very much for beginners but if you're just starting to find out all about using Lisp to do web development this might be a good read:
http://www.gigamonkeys.com/book/practical-web-programming-with-allegroserve.html
http://www.gigamonkeys.com/book/practical-web-programming-with-allegroserve.html
Thank you for the link. I used to code in Scheme and enjoyed it very much, it would be great to get back into LISP one of these days. However, whenever I see HTML generated from within the code, as it is in that tutorial, it just feels like it is 10 years behind the state of the art. I think it is much better to use some kind of HTML templates (but maybe I am wrong?).
Another thing I was looking for when I last evaluated LISP was an OR-mapping library. I can't go back to executing SQL statements from within my code - that is also so 90ies. I think there is one project underway to provide OR-mapping for LISP, but I am not sure if it was ready for prime time yet.
Another thing I was looking for when I last evaluated LISP was an OR-mapping library. I can't go back to executing SQL statements from within my code - that is also so 90ies. I think there is one project underway to provide OR-mapping for LISP, but I am not sure if it was ready for prime time yet.
It's possible, and reasonably common to use templates with Lisp web apps. HTML-TEMPLATE and Tal (from Zope) are widely used.
HTML-TEMPLATE: http://weitz.de/html-template/
Tal support in yaclml, part of Uncommonweb: http://common-lisp.net/project/bese/yaclml.html
CLSQL provides object/relational mapping; it's quite good: http://clsql.b9.com/
HTML-TEMPLATE: http://weitz.de/html-template/
Tal support in yaclml, part of Uncommonweb: http://common-lisp.net/project/bese/yaclml.html
CLSQL provides object/relational mapping; it's quite good: http://clsql.b9.com/
Not exactly what you asked for, but another option is to use persistent objects (e.g. AllegroCache) instead of mapping them to RDBMS.
Please pardon how newbieish this article starts out. I don't mean to insinuate that you're at this level - I was just dropping a link that may or not may be of use to you.
I'm a newbie and this is really helpful. I went through Uncommon Web's doc and it was too big of a jump for me. PCL seems pretty thorough and concise for my level.
Just curious, after this, in what order should I learn things? It seems that most CL libraries have very sparse docs. Does that mean I should develop a habit of reading the source code of everything I learn?
Just curious, after this, in what order should I learn things? It seems that most CL libraries have very sparse docs. Does that mean I should develop a habit of reading the source code of everything I learn?
Wow, what a response. I am excited by the prospect that at some point in the near future I can dive into lisp with a clear goal in mind and the resources to make it happen. I have been hearing for years about how lisp is really the ultimate programmers language. I respect this lisp-lover's oppinion and experience ultimately, and put him right in the list with Paul and Carl. So, 3 elite programmers who sing the graces of lisp. One was enough for me to want to learn to think in lisp. 3 in support should be a point to ponder.
The problem I have always encountered with trying to build a solution in lisp is the lack of resources. It sounds like lisp is inching closer and closer to the critical mass where tools and libraries will finally be available for more then a few uses.
This lisper I know said "Don't hold your breath." when I suggested that lisp may be coming close to getting its due. I know that there are alot of people out there like me. We will start using lisp when the great advantages of the language finally outweigh the disadvantages of implementation. I think its just a matter of time until lisp is a clear win, as long as the tools come.
I've never written a line of lisp. I want to save my lisp cherry for a time when I can have more "Yay" then "Doh" while I learn and use it. I think it will be worth the wait.
The problem I have always encountered with trying to build a solution in lisp is the lack of resources. It sounds like lisp is inching closer and closer to the critical mass where tools and libraries will finally be available for more then a few uses.
This lisper I know said "Don't hold your breath." when I suggested that lisp may be coming close to getting its due. I know that there are alot of people out there like me. We will start using lisp when the great advantages of the language finally outweigh the disadvantages of implementation. I think its just a matter of time until lisp is a clear win, as long as the tools come.
I've never written a line of lisp. I want to save my lisp cherry for a time when I can have more "Yay" then "Doh" while I learn and use it. I think it will be worth the wait.
From my limited perspective, it looks like one of the major drawbacks (evidenced in this discussion) is the lack of critical mass for one specific Lisp stack. Having so many competing setups makes for:
-little documentation
-little support
-limited libraries
-super tiny communities
Compare that to LAMP(P = PHP, Perl, or Python) and Rails. Their communities, docs, support, libs, etc are HUGE compared to the tiny fractured Lisp ecosystem.
Do LAMP and Rails have their own problems? Of course. But the point is that they have been proven over and over and over. Lisp is certainly sexy and will give you that toned body and hairy chest you always wanted, but beyond Viaweb and Reddit (cough, cough) can you readily (and without searching) think of other top sites that have delivered a major hit with it? Second, can you think of any hit Lisp sites that didn't involve Paul Graham? I rest my case.
I'd love to see this change, but unfortunately that's the current reality.
------
Update: After doing some searching, I also found ITA's backend Orbitz software written in Lisp. That certainly was a hit and is also used by many other systems. The author is Carl de Marcken and this hit is also mentioned in pg's "Great Hackers" essay.
So, Two Lisp Hit-makers:
Paul Graham, Phd in CS from Harvard
Carl de Marcken, Phd in CS from MIT
So, if you're a Phd in CS from one of the top programs in the world - I'd say Go For It! ;)
-little documentation
-little support
-limited libraries
-super tiny communities
Compare that to LAMP(P = PHP, Perl, or Python) and Rails. Their communities, docs, support, libs, etc are HUGE compared to the tiny fractured Lisp ecosystem.
Do LAMP and Rails have their own problems? Of course. But the point is that they have been proven over and over and over. Lisp is certainly sexy and will give you that toned body and hairy chest you always wanted, but beyond Viaweb and Reddit (cough, cough) can you readily (and without searching) think of other top sites that have delivered a major hit with it? Second, can you think of any hit Lisp sites that didn't involve Paul Graham? I rest my case.
I'd love to see this change, but unfortunately that's the current reality.
------
Update: After doing some searching, I also found ITA's backend Orbitz software written in Lisp. That certainly was a hit and is also used by many other systems. The author is Carl de Marcken and this hit is also mentioned in pg's "Great Hackers" essay.
So, Two Lisp Hit-makers:
Paul Graham, Phd in CS from Harvard
Carl de Marcken, Phd in CS from MIT
So, if you're a Phd in CS from one of the top programs in the world - I'd say Go For It! ;)
Sorry, I spent years demonstrating to people how Lisp is better for everybody doing web development, not just elite programmers. I'm done. My new philosophy is, if you can't get them to join you, beat them.
Now I'm at a point where my venture absolutely must succeed. I'll take every advantage I can get.
Now I'm at a point where my venture absolutely must succeed. I'll take every advantage I can get.
It's scheme, not CL, but don't forget HOP http://hop.inria.fr/ and whatever the PLT people have in store http://www.plt-scheme.org/
btw, HOP uses the Bigaloo web server and so isn't quite adequate apparently for a high traffic site (like I assume most here will have). here's an email exchange I had with the author the other day- sorry if the comment's a bit big:
I apologize for such a late answer. I'm always fighting with email. I'm definitively not good at it because I'm not able to answer at a decent pace.
Hi Manuel. I've just been playing with Hop. Well done. It's just beautiful. A couple of questions about it: first, the bigloo web server- is this completely necessary? I guess one of 4 things would need to happen for hop to be totally ideal: Bigloo implements light-weight threads Gambit-scheme style so that it can have erlang-like load capacity ( http://www.sics.se/~joe/apachevsyaws.html ) (most preferable option, IMO, though I have no idea what your feelings are here. To me this would make bigloo light-years ahead of all other scheme imps.) I have tried several things concerning the multi-threading. You can read our FairThreads proposal (http://www.inria.fr/mimosa/Manuel.Serrano). When I did this study I though that FairThreads was a decent solution. Now that's no longer clear to me. Hop does not use Fair threads but Posix like threads. I'm quite aware of all the troubles that come with this kind of threads. I have not considered an Erlang-like solution yet but that's something I could envision.
Get hop to work with a gambit-scheme server instead of bigloo (where would i start with this? can you point me to relevant points in the code?) No, I think that you cannot. Hop is really not a portable Scheme program. It extensively relies on Bigloo features. I'm afraid that you will have to rewrite entirely. It's not a huge program but it is not that small either. The problem is that it uses a lot of libraries that you would have to re-implement too. All in all it's about 40.000 lines of Scheme code. I don't really recommend to try to re-implement that (except if you are very brave ;-)
Get hop to work with mod_lisp (less preferable, IMO) No. I don't think that a solution too. Hop needs to be running permanently. That is two successive requests must be handled in the same execution environment (i.e., the first request may set a global variable that is read by the second request). Hence, even if you want to use Apache you will be forced to have a server running somewhere and you will have to establish a communication between Apache and that server. Why not directly using Hop as a server?
Second item is the overall latency and possible redundancy... actually, I don't know how to phrase the problem- but it boils down to this, I counted over 27 distinct "connecting to server" cycles when you click on the "weblets" tab of the Hop Home page. It took quite a while re-establishing the connection to the server 27 times. Are there thoughts on overcoming this? Not really but... First, the Hop server supports keep-alive connections. This should help. Then, honestly, I have hoped to find time to optimize the server for months. I have not found that time yet but I think that I will have it around the end of Spring. At that moment, I will study the problem you are mentioning.
However, I'm using Hop on a daily basis. Even more, in order to push it to its limits, I'm using it as my regular proxy. That is all my HTTP connections, on all my machines, go through a Hop proxy/server. The performance are acceptable. That is, I'm not annoyed by an important slow down due to Hop.
Thanks again!! You are welcome. Thanks for the encouragements.
-- Manuel
I apologize for such a late answer. I'm always fighting with email. I'm definitively not good at it because I'm not able to answer at a decent pace.
Hi Manuel. I've just been playing with Hop. Well done. It's just beautiful. A couple of questions about it: first, the bigloo web server- is this completely necessary? I guess one of 4 things would need to happen for hop to be totally ideal: Bigloo implements light-weight threads Gambit-scheme style so that it can have erlang-like load capacity ( http://www.sics.se/~joe/apachevsyaws.html ) (most preferable option, IMO, though I have no idea what your feelings are here. To me this would make bigloo light-years ahead of all other scheme imps.) I have tried several things concerning the multi-threading. You can read our FairThreads proposal (http://www.inria.fr/mimosa/Manuel.Serrano). When I did this study I though that FairThreads was a decent solution. Now that's no longer clear to me. Hop does not use Fair threads but Posix like threads. I'm quite aware of all the troubles that come with this kind of threads. I have not considered an Erlang-like solution yet but that's something I could envision.
Get hop to work with a gambit-scheme server instead of bigloo (where would i start with this? can you point me to relevant points in the code?) No, I think that you cannot. Hop is really not a portable Scheme program. It extensively relies on Bigloo features. I'm afraid that you will have to rewrite entirely. It's not a huge program but it is not that small either. The problem is that it uses a lot of libraries that you would have to re-implement too. All in all it's about 40.000 lines of Scheme code. I don't really recommend to try to re-implement that (except if you are very brave ;-)
Get hop to work with mod_lisp (less preferable, IMO) No. I don't think that a solution too. Hop needs to be running permanently. That is two successive requests must be handled in the same execution environment (i.e., the first request may set a global variable that is read by the second request). Hence, even if you want to use Apache you will be forced to have a server running somewhere and you will have to establish a communication between Apache and that server. Why not directly using Hop as a server?
Second item is the overall latency and possible redundancy... actually, I don't know how to phrase the problem- but it boils down to this, I counted over 27 distinct "connecting to server" cycles when you click on the "weblets" tab of the Hop Home page. It took quite a while re-establishing the connection to the server 27 times. Are there thoughts on overcoming this? Not really but... First, the Hop server supports keep-alive connections. This should help. Then, honestly, I have hoped to find time to optimize the server for months. I have not found that time yet but I think that I will have it around the end of Spring. At that moment, I will study the problem you are mentioning.
However, I'm using Hop on a daily basis. Even more, in order to push it to its limits, I'm using it as my regular proxy. That is all my HTTP connections, on all my machines, go through a Hop proxy/server. The performance are acceptable. That is, I'm not annoyed by an important slow down due to Hop.
Thanks again!! You are welcome. Thanks for the encouragements.
-- Manuel
if you're thinking of developing a web app in lisp but haven't looked at python or ruby, you might consider them -- remember, it's not just the raw language but also the surrounding infrastructure, and python (django, turbogears) and ruby (rails) have arguably much more developed frameworks, and also arguably have most of the interesting language features that would draw you to lisp in the first place.
the point is the language is only one piece of the equation -- you want to be spending your time developing your app, not rolling your own supporting elements (templating systems, init scripts, form validators, sql bindings/mappers, etc.) and of course i'm sure others will point out reddit's switch from lisp to python.
the point is the language is only one piece of the equation -- you want to be spending your time developing your app, not rolling your own supporting elements (templating systems, init scripts, form validators, sql bindings/mappers, etc.) and of course i'm sure others will point out reddit's switch from lisp to python.
Most people say that you shouldn't use lisp because it doesn't have libraries. Well they are only half right.
You shouldn't use lisp, because most lisp libraries have
very little documentation. Lisp has a lot of libraries.
You have to be willing to read code (i.e. examples/tests)
to figure out how to use them.
http://www.cliki.net/Library
Assuming that the most precious commodity is development time. I'd say rails/django don't even come close to UCW in getting stuff done.
http://www.cliki.net/Library
Assuming that the most precious commodity is development time. I'd say rails/django don't even come close to UCW in getting stuff done.
Why is lisp so sparsely documented?
For me saying what a function does isn't good enough, when one is documenting an api. Examples are a must. The other problem is that the lisp packaging system is a bit weird. If a library doesn't have an ASDF definition its almost non-existent for me.
(because-function-names-are-so-long-and-detailed 'post)
For our stuff, I started with rails, just found it too hard to get started. I had to decide what the tables looked like, how I was going to divide the data, write a model for every table then connect it to a page.
With UCW, I write code which connects to a function call. Simple.