3 ms·

Well now I feel old. I remember writing a web app in Tcl way back in 2000 or so right after I graduated. From what I recall, AOLServer was not as fun to program

Well now I feel old. I remember writing a web app in Tcl way back in 2000 or so right after I graduated. From what I recall, AOLServer was not as fun to program as this appears to be. http://aolserver.github.io/ http://aolserver.github.io/ TCL was actually a pretty OK language. I do remember that aolserver was a major improvement over CGI scripts.

mst 12y ago
Relaying for xan: It still is =) And I hope to spread that love as much as possible by carefully cultivating a tiny, cute little ecosystem of lovely tools on top of tänzer, and help foster the idea of making wonderful web framework middlewares for Tcl to make it a much easier proposition to proliferate our wares to the masses.
pjmlp 12y ago
I joined a Portuguese startup in 1999 that had an in-house Apache module for TCL, an application server similar to AOLServer in design. With an abstraction layer similar in capabilities to what people know Rails for. With lots of C plugins as well. Being part of the core developers meant I was a regular reader on tcl/tk forums, eagerly looking for the upcoming TIPs, following ActiveState work in the community and so on. It was also the last time in my career that I used C on the job, instead of my then favourite C++. I learned lots of stuff while there, but two important things in terms of programming languages were: 1 - scripting languages are great for prototyping 2 - if you ever plan to scale, use a compiled language for the get go instead (JIT/AOT) On our case, due to certain circumstances, we were part of the .NET alpha program and started moving part of architecture to C#/Managed C++.
- davidw 12y ago
  > 2 - if you ever plan to scale, use a compiled language for the get go instead (JIT/AOT) Like Facebook? I don't think that's good advice, for several reasons: * Everyone dreams about scaling. For most, it's just not a pertinent reality. It's easy to prematurely optimize. * Scripting languages will let you iterate faster, and find some kind of product-market fit. Think about Twitter: I love Rails, but ... no, Rails is not a good fit for what Twitter is. And yet they managed to right the ship and move on to more appropriate systems. If you've got a viable business, you can find the money to improve.
  - pjmlp 12y ago
    Many of the improvements Facebook has brought into PHP eco-system is because the code base is so big, that it is worthwhile to invest so much money into research about PHP native code generation that re-write the whole stuff. I get your point, but there are many languages that offer the rapid iteration of dynamic languages, while offering native code. And monkey patching just doesn't scale across teams.
xantronix 12y ago
Don't feel old! Tcl is only two years younger than me and three years older than Python, and hardly anyone considers Python a relic. One day I shall adequately understand why it is that Tcl fell out of favor. The core language was never bad nor yucky, but things did improve quite a bit when Tcl 8.0 came along, with internal typed representations and whatnot. However, I finally decided this was the year I "put up or shut up" and prove to myself an interesting theorem that language adoption is mostly marketing. Here's to hoping the quirky little devil known as Tcl shall usher in a newfound hipster resurgence of joyful web software!

mst 12y ago

Relaying for xan: It still is =) And I hope to spread that love as much as possible by carefully cultivating a tiny, cute little ecosystem of lovely tools on top of tänzer, and help foster the idea of making wonderful web framework middlewares for Tcl to make it a much easier proposition to proliferate our wares to the masses.

pjmlp 12y ago

I joined a Portuguese startup in 1999 that had an in-house Apache module for TCL, an application server similar to AOLServer in design. With an abstraction layer similar in capabilities to what people know Rails for. With lots of C plugins as well. Being part of the core developers meant I was a regular reader on tcl/tk forums, eagerly looking for the upcoming TIPs, following ActiveState work in the community and so on. It was also the last time in my career that I used C on the job, instead of my then favourite C++. I learned lots of stuff while there, but two important things in terms of programming languages were: 1 - scripting languages are great for prototyping 2 - if you ever plan to scale, use a compiled language for the get go instead (JIT/AOT) On our case, due to certain circumstances, we were part of the .NET alpha program and started moving part of architecture to C#/Managed C++.

davidw 12y ago
> 2 - if you ever plan to scale, use a compiled language for the get go instead (JIT/AOT) Like Facebook? I don't think that's good advice, for several reasons: * Everyone dreams about scaling. For most, it's just not a pertinent reality. It's easy to prematurely optimize. * Scripting languages will let you iterate faster, and find some kind of product-market fit. Think about Twitter: I love Rails, but ... no, Rails is not a good fit for what Twitter is. And yet they managed to right the ship and move on to more appropriate systems. If you've got a viable business, you can find the money to improve.
- pjmlp 12y ago
  Many of the improvements Facebook has brought into PHP eco-system is because the code base is so big, that it is worthwhile to invest so much money into research about PHP native code generation that re-write the whole stuff. I get your point, but there are many languages that offer the rapid iteration of dynamic languages, while offering native code. And monkey patching just doesn't scale across teams.

xantronix 12y ago

Don't feel old! Tcl is only two years younger than me and three years older than Python, and hardly anyone considers Python a relic. One day I shall adequately understand why it is that Tcl fell out of favor. The core language was never bad nor yucky, but things did improve quite a bit when Tcl 8.0 came along, with internal typed representations and whatnot. However, I finally decided this was the year I "put up or shut up" and prove to myself an interesting theorem that language adoption is mostly marketing. Here's to hoping the quirky little devil known as Tcl shall usher in a newfound hipster resurgence of joyful web software!