A view from the hill - Comments

Matt: Fun with loop

nospam@example.com (Matt) — Wed, 11 Jun 2008 18:12:02 +0200

OK. It's a very old post, but for the record let me
give a version that's closer to the core of CL:
(let ((list '(a b c d e f g h i j k l m))
(limit 5))
(do ((rest list (rest rest))
(acc nil (push (first rest) acc))
(count 0 (1+ count)))
((or (>= count limit)
(null rest))
(values (reverse acc)
rest))))

For keeping track of first, rest, and index during iteration, do works just fine.

The main benefit of your version (probably only important for very large values of "limit") is that collect in loop or
iterate might avoid the final reverse operation.

Friedrich: UCW leadership change

nospam@example.com (Friedrich) — Thu, 01 May 2008 07:35:54 +0200

Ok, well I have a harsh dispute from Drew because of his lisp-on-lines. which I judged as unusable.
Well it is based on UCW for which I wrote at least some tutorial which has worked at least up to 2007 somewhat. I did not received any feedback on it neither be it by email or in the Wiki where I wrot e it. I for my part can just give very sceptical juddgment for UCW. It's complex, hard to understand and it's unclear on how generated code, templated etc are working together. I spend months on getting somewhere with it with not that much success.

And we now do have alternatives which are IMHO
1) easier to understand
2) easier to setup
3) easier to get somewhere with it

As there are
1) AllegoServe + Webactions (very "traditional" and so what you learned elswhere can be applied directly")
2) the Edi-ware based web development stuff (it's also more traditional)

I put that much time into UCW because I thought it was the right thing. I'm not that convinced of that any longer....

Regards
Friedrich

attila lendvai: Escaping from sql-reader-syntax in CL-SQL

nospam@example.com (attila lendvai) — Mon, 25 Feb 2008 04:28:39 +0100

i've annotated Marijn's paste with some syntax examples from the cl-rdbms testsuite.

Marijn Haverbeke: Escaping from sql-reader-syntax in CL-SQL

nospam@example.com (Marijn Haverbeke) — Sat, 23 Feb 2008 10:52:59 +0100

Just to show off, see http://paste.lisp.org/display/56365 for the Postmodern variant. Also not that Postmodern will actually let you nest queries if you need to

Sasha Kovar: Twentieth century boys

nospam@example.com (Sasha Kovar) — Wed, 23 Jan 2008 20:03:39 +0100

A quick data point: I've got a production UCW app running that's been getting 10-20k pageviews/day without breaking a sweat on a pretty modest virtual server (linode). Hopefully soon it'll be seeing much more traffic

I like cl-migrations and writing my db code in sql. Much simpler than translating/abstracting, as Sohail points out. I recommend pg-introspect as well, as it keeps all db definitions in the one place it should be. I've never seen an ORM layer handle the subtleties of db constraints well. Something is always lost in the abstraction. Why bother? That code belongs in the db.

Sorry, no SSL or shopping cart, but managing sessions and doing input validation under UCW are simple. I love the components and continuations model, although it took me a little while to get acclimated. I'm only using the continuation style with about half of the toplevel window components. UCW is nice in that you can choose to use the hunchentoot/traditional/RESTy style when that makes more sense.

Haven't needed to split to multiple servers yet. Should be easy though: second db machine, and if we need to run multiple UCWs just sticky-session load balance behind a proxy. I'm not a fan of keeping session state in the db, but then losing some session data on crash isn't a huge deal for me.

I'm interested in watching how weblocks develops. It's got a lot of potential.

Leslie P. Polzer: Twentieth century boys

nospam@example.com (Leslie P. Polzer) — Wed, 23 Jan 2008 10:21:48 +0100

I'm developing a browser game with Weblocks.
SSL is not of importance here, and distributed computing isn't an issue yet, but all the other things you might want to have in a Web 2.0 app are there.

So far it has worked out fine and with a huge lot less hassle than I had had if I wrote the thing in, yuk, plain PHP (or frameworked PHP, for that matter).

I suggest you try out Weblocks and see for yourself. It's easy to get started (although continuing is a bit more involved, as it's a quite young framework).

Leslie

anonymous: Twentieth century boys

nospam@example.com (anonymous) — Wed, 23 Jan 2008 00:45:44 +0100

While not lisp, www.dabbledb.com was written in Seaside.

ken: Twentieth century boys

nospam@example.com (ken) — Wed, 23 Jan 2008 00:30:02 +0100

I don't see a real problem with CL-MIGRATIONS requiring you to write SQL. With any non-trivial problem in ActiveRecord, you need to write SQL, too. It just happens to be in Ruby syntax. (Good luck figuring out what to pass to :joins!) I always have to haul out my big SQL book, write and debug SQL, and then convert it to Ruby. And if you ever change databases (I've done it!), you'll need to debug a bunch of SQL, anyway.

Rails has many problems (don't get me started -- heh!), but an "emphasis on wizard-like code generation" is not one of them. I don't think any of my current Rails project is generated code, and that seems to be the concensus among the other programmers I know. There is a scaffolding system, but if the program lives longer than about a week, it tends to all get replaced. That's what scaffoldings are.

Sohail: Twentieth century boys

nospam@example.com (Sohail) — Tue, 22 Jan 2008 22:29:24 +0100

Hey, nice post!

I think the thing is that Rails on Lisp actually exists. It is just tied up in a bunch of libraries. My favourite features of Rails are the ORM, MVC and migrations, which are adaquetly implemented by CLSQL, either HTML-TEMPLATE or CL-WHO (I prefer CL-WHO but purists would prefer HTML-TEMPLATE) and CL-MIGRATIONS. The only problem with CL-MIGRATIONS is that it requires you to write your own SQL, but not a big deal.

In the end, I found after reading the Agile Web Dev book that there isn't anything that extra special to Rails that isn't served by existing Lisp libraries.

The thing I despise about rails is the emphasis on wizard-like code generation. My view on code generation is that you either generate the code every time (like Lisp macros) or you don't generate it at all. It is stupid, stupid, stupid, to generate code once.

That being said, Rails fulfills a large niche: web-enabled databases. It is dumb easy to expose a database through the browser with Rails.

On the other hand, Weblocks does things a lot differently in that it focuses on components. I'm still up in the air about the current design of the system as overriding the generated HTML is not very intuitive, but the idea is solid.

Olaf: Swing the heartache: Interactive DB maintenance in Lisp

nospam@example.com (Olaf) — Sun, 13 Jan 2008 20:22:51 +0100

Why actually fetch the data? Either use alter table sometable alter column data ...; Or create a new column, then update the new column with the data from the old column (update sometable set newdata=data) and finally drop the data column and rename newdata to data. Much more efficient and just as elegant.

Andi: Swing the heartache: Interactive DB maintenance in Lisp

nospam@example.com (Andi) — Thu, 22 Nov 2007 16:23:02 +0100

thanks for your kind answer!

Holger Schauer: Swing the heartache: Interactive DB maintenance in Lisp

nospam@example.com (Holger Schauer) — Thu, 22 Nov 2007 15:57:15 +0100

It refers to a song of the same name by the band "Bauhaus", which some might claim have established the gothic genre. I don't know if the title also names a band.

Apart from that I like using song references as blog titles, I found dealing with the db issue in the cl repl to be a swinging solution to a particular headache, so that fitted quite nicely.

Andi: Swing the heartache: Interactive DB maintenance in Lisp

nospam@example.com (Andi) — Thu, 22 Nov 2007 12:01:20 +0100

Holger,

what means "swing the heartache"? Does it refer to the goth band?

Holger Schauer: Lisp golf

nospam@example.com (Holger Schauer) — Sun, 11 Nov 2007 14:57:49 +0100

This one doesn't count as it modifies the input. A similar approach is doable in CL with the same RE library (cl-ppcre) as in the blog post.

Holger Schauer: Lisp golf

nospam@example.com (Holger Schauer) — Sun, 11 Nov 2007 14:55:51 +0100

Lisp Users Who Actually Like Format And Loop Damnit -- LUWALFALD. I like that. And if you add the C in front for Club, you actually get the usual CL- prefix.

BTW: It's not like I dislike format or loop. It's more a chicken and egg-problem: I don't use them regularly, so I sometimes find solutions that use them harder to understand than others involving, say, dolist, which I do use regularly. So my comment in the blog post says a lot more about me than about format/loop.