Monday, August 08, 2005

The Perfect Python IDE

There are a lot of things I look for in a good IDE. Looking for those things, I've tried Kate, KDevelop, Eric3, Komodo, and Boa-Constructor, to name only a few. Some of them give me a few things I like, each one giving different things. But, none of them have all of the core features I really want in a Python editor.
So, here I am laying them out. Maybe an IDE will be developed meeting these requests, or maybe I'll do it myself eventually, possibly joining one of the above mentioned projects.
Such an imaginary program we will call, for the time being, The Perfect Python IDE.
  • Indentation is important in Python programs. The fact that one can not mix tabs and spaces brings hardship to anyone who commonly edits multiple projects' files simultaneously, when they use different indenting rules. The Perfect Python IDE sees what kind of indenting is being used in the current file, and auto-indent following that style.
  • indentation exists both in the actual code and in how the code is displayed. Sometimes a line may be lengthy on a particular display, or when I'm using a horizontally split view. The Perfect Python IDE dynamically wraps the line's display without affecting the source code itself, and even indents the following line for clarity.
  • Of course, The Perfect Python IDE knows that if any feature gets in the way, its not a feature; it is a bug. That's why it only follows through with an auto-completion when I explicitly tell it to, and it always implicitly goes away the moment it sees I don't want auto-completion.
  • Even the smallest of projects needs a sense of history. You never know when you'll want to undo something you saved a week ago, so version control is a must. An IDE should naturally compliment such a system, like CVS or Subversion. The Perfect Python IDE only asks me for a path to the local working copy, or a svn+ssh:// URL to the repository, so it can just keep the working copy files hidden while I work on them.
  • As long as I'm just giving The Perfect Python IDE the URL to the repository, there isn't a need for any silly project files.
  • Now that my files are stored in a repository on a remote server, interested people might work on the code, too. They'll probably want the code to work when they update their working copies, so I'd better test before every time I commit something. The Perfect Python IDE can see my test/ directory and my test_*.py scripts and run them, being sure to set the PYTHONPATH for my package.
  • UnitTests being run is one thing, but doing something with them when the tests fail is something to be really proud of. The Perfect Python IDE extracts the tracebacks and bookmarks all the lines in the call stack. It even numbers them and displays local values at the time of error.
  • The Perfect Python IDE may be perfect, but part of that is being exactly what I want it to be, or what you want it to be. Those things may be different, but The Perfect Python IDE can be different, too. It can be customized at any point; and, it can be customized in Python. Of course, it is used as an editor of those very same customization scripts.
  • Not everything needs custom code to make it act right. The settings and configuration can go a long way, too. The Perfect Python IDE never asks me to restart it for settings to take affect, and it always lets me know exactly what any particular setting does.
Do I ask too much? I could probably ask for more. Did I not specify well enough what I want? I'm mostly listing it to get things straight in my own head. Does something already exist that does this all? I haven't it. Do Vim or (X)Emacs do the things I ask? They're ugly. I pay for high-end computers and large monitors for graphical interfaces, not consoles.
Yeah, they both have GUI modes, but they kind of suck. They're little more than terminal emulators enhanced for that particular program. Either of them could be a good GUI IDE, but communities around them just don't seem to understand the reason to make it looks "pretty", so I doubt they will, at least, not for a very long time. Despite what many techno-machos say, programs really are better when they're pretty.

2 comments:

Straw Dogs Python Blog said...

Personally I'm a big fan of Komodo. I recently did a review of Python IDE's and that came out on top. Its also developing quite fast and hgas a massive support and community behind it.

If you were interested in seeing my review then visit my blog.

BLANCHE said...

Nice blog. Have you seen your google rating? BlogFlux It's Free and you can add a Little Script to your site that will tell everyone your ranking. I think yours was a 3. I guess you'll have to check it out.

Computer News
China's Google rockets on debut
It was a remarkable debut. Chinese search engine Baidu.com's shares sold in the United States for US$27 ($39) - then surged on day one to close at US$122.54 ($177.46).

Friday's result was the biggest first-day gain for a new listing in the US for five years.

Investors had more than quadrupled their money.

An analyst in New York with IPOdesktop.com, a website devoted to initial public offerings, John Fitzgibbon, said: "This one is the return to the internet bubble. Last time we saw a deal skyrocket was during the frothy IPO markets of 1999 and 2000."

Then came the post-mortem.

Some analysts said the internet search engine could have had an even better payday for itself if underwriters had sold the deal at a higher price to begin with.

"It looks to me like the underwriters should have had a better indication of the appetite for this stock than they did," said Donald Straszheim, president of Straszheim Global Advisors.

Investors were eager to own a stake in a company that many say could grow as dramatically as Google and is based in a country itself undergoing explosive growth.

But other analysts are not so sure the underwriters made a mistake, given that the company's shares were priced at a relatively high multiple of revenues.

"It wasn't priced out of line with the rest of the market. In a situation like this, you're dealing with the unpredictability of the after market," said Tom Taulli, of Instream Partners in Newport Beach, California.

At issue is the underwriting process. Banks selling shares to the public - in this case Credit Suisse First Boston, Goldman Sachs, and Piper Jaffray - are paid to use quantitative models to determine a fair price for shares, but also to gauge investor demand.

Underwriting has both subjective and objective elements, making definitive evaluation of a bank's performance difficult. In this case, demand for the IPO was evidently outsized, but so were the unknowns for the company.

"We don't know the potential impact of censorship in China, or how quickly the internet will grow there," said David Menlow, president of IPOfinancial.com. Had the IPO been priced higher and then fallen in the first day of trading, investors could have sued.

Baidu.com chairman and chief executive Robin Li, speaking on CNBC, said he was not upset by the potential lost proceeds of the IPO because the company had only sold a small portion of itself, and had significantly more growth ahead.

But University of Florida Professor Jay Ritter, an IPO expert, said Baidu.com left money on the table by introducing the shares at a level well below where they ended hours later.


Copyright © - 2005 Entireweb

=============================================
Rippers