Python vs PHP

[Catalin]

Can Python replace PHP?
Can I use a python program to make an interface to a mysql 4.X database?
If that's possible where can I find a tutorial?

 

[News M Claveau /Hamster-P]

Hi !

PHP-5 let down MySql, for Sql-Lite (rumor ?)

Python has good support for MySql AND Sql-Lite.

@+

Michel Claveau

 

[Aahz]

Can Python replace PHP?
Can I use a python program to make an interface to a mysql 4.X database?

The short answer is "yes" to both questions, but with a caveat: Python is
a general-purpose programming language, where PHP is designed to write
web applications. So it's not quite as simple/straightforward to use
Python the same way you use PHP for plain web pages; conversely, if
you've been running into difficulty with PHP in trying to write
complicated program logic, you'll find Python a big help.

 

[Bruno Desthuilliers]

Hi !

PHP-5 let down MySql, for Sql-Lite (rumor ?)

PHP 5 won't come bundled with a default MySql lib, but anyone is still
free to link against the official MySQL lib, one way or another...

And yes, the default 'bundled-in' SQL lib should be Sql-lite.

Bruno

 

[Jaroslaw Zabiello]

Can Python replace PHP?

Sure. Look at http://spyce.sourceforge.net

Can I use a python program to make an interface to a mysql 4.X database?

Of course, you can. http://www.python.org/sigs/db-sig/
E.g. http://dustman.net/andy/python/MySQLdb_obsolete

If that's possible where can I find a tutorial?

http://www.python.org/doc/Newbies.html
http://www.python.org/doc/current/
http://www.python.org/cgi-bin/faqw.py

 

[Jp Calderone]

Sure. Look at http://spyce.sourceforge.net

Lest the OP think Spyce is the only or even the preferred solution for web
programming with Python:

http://cherrypy.org/

http://www.twistedmatrix.com/

http://skunkweb.sf.net

http://webware.sf.net

http://zope.org

http://www.mems-exchange.org/software/quixote/

http://www.object-craft.com.au/projects/albatross

http://jonpy.sf.net

http://jotweb.tummy.com/

And let's not forget plain old CGI and mod_python.

Jp

 

[Afanasiy]

Can Python replace PHP?
Can I use a python program to make an interface to a mysql 4.X database?
If that's possible where can I find a tutorial?

I recommend Spyce, Webware, or Quixote depending on your requirements,
but Draco, SkunkWeb, RoadKill are worth looking into as well. Spyce is
really the best (for PHP-ness) in my opinion, after you actually get past
the fact that it supports your old fashioned notion of what embedded
delimiters should look like, as well as a new CTS-friendly alternative.

Spyce is the closest thing to PHP, but also provides some features PHP
does not. However, I think you will find some people, like myself, who are
still unable to make the switch from PHP to Python for web development,
because of a lot of little things. These people know Python is better and
may even know their non-web PHP applications are hell compared to their
Python re-implementations. However, they still use PHP for web development
and feel they have to. They may have even written out all the requirements
and received vague responses to them.

Python web development is fragmented. Mod_python CVS now contains the work
of what used to be mod_psp by Sterling Hughes, a PHP warlord, I'm sure
he'll reply. This possibly soon to be official 'PSP' is not much in my
opinion and should probably not be folded into mod_python. Others agree:

http://www.modpython.org/pipermail/mod_python/2003-June/003200.html

Another big problem with Python web development is the lack of good shared
hosting support. Primarily the lack of anything resembling PHP's safe
mode. Mod_python has no equivalent of this safe mode, but Apache 2's
'Perchild MPM' is supposed deprecates this need. However, this still
requires an httpd.conf edit and a restart of Apache for every user added,
very annoying, especially to a shared hosting service.

http://apache.slashdot.org/apache/02/12/02/1346206.shtml?tid=
http://httpd.apache.org/docs-2.0/mod/perchild.html
http://www.modpython.org/pipermail/mod_python/2001-November/001705.html

A couple other slight issues :

* catch-all error handling not as flexible or non-existent
* no equivalent to php_value auto_prepend_file "_header.php"
* no equivalent to php_value auto_append_file "_footer.php"
* not as intelligent form value handling
* white space sensitivity is a unique problem when embedding

The best options for me are currently :

* use PHP
* use Spyce
* fork Spyce (creating more fragmentation)
* embed python straight into my favorite platform (quite funny)
* write my own mod_python-based platform (more fragmentation)
* find the holy grail (i must have missed it)

I've been handling this dilemma for months. I've even benchmarked quite a
few possible solutions, with pleasing results in those which I thought
were well done, and pleasing results for Python over PHP in general.

-AB

 

[Jon Ribbens]

A couple other slight issues :

* catch-all error handling not as flexible or non-existent
* no equivalent to php_value auto_prepend_file "_header.php"
* no equivalent to php_value auto_append_file "_footer.php"
* not as intelligent form value handling
* white space sensitivity is a unique problem when embedding

Try jonpy (http://jonpy.sf.net/), it solves all of the above.
Also you may find FastCGI (which jonpy supports) helps with your
"safe mode" problem, although it might require a bit of hackery.

 

[Afanasiy]

A couple other slight issues :

* catch-all error handling not as flexible or non-existent
* no equivalent to php_value auto_prepend_file "_header.php"
* no equivalent to php_value auto_append_file "_footer.php"
* not as intelligent form value handling
* white space sensitivity is a unique problem when embedding

I think I should add import, pythonpath and module caching to this list.
These are much more annoying than PHP's inclusion mechanisms. I am even
annoyed in non-web applications by them quite often, but not necessarily
all at the same time.

Also add to the list, the need for PyChecker. You must run PyChecker
to discover errors which would occur on a section of logic being hit
which contained a silly typographical error. PHP would discover these
without them being hit, when you were developing the page.

Python is more beautiful and more powerful, but for this purpose I
consider the current implementations still lacking compared to PHP.
So much so that I am unable to bring myself to write new PHP, but
unable to continue using Python until I find the perfect implementation.

-AB

 

[Afanasiy]

Try jonpy (http://jonpy.sf.net/), it solves all of the above.
Also you may find FastCGI (which jonpy supports) helps with your
"safe mode" problem, although it might require a bit of hackery.

No, CGI is not an option and I tried jonpy. I kept my notes about it...

 

[Jon Ribbens]

CGI, FastCGI, SCGI are not faster than mod_python in my experience and
straightforward benchmark.

In the case of CGI, yes of course. In the case of FastCGI, your
experience contradicts my own.

That's all I can really say about it. CGI is not an option for me.

I don't understand why you keep saying "CGI is not an option" when
nobody has suggested that you use it.

 

[Afanasiy]

I don't understand why you keep saying "CGI is not an option" when
nobody has suggested that you use it.

Ok

 

[A.M. Kuchling]

I don't understand why you keep saying "CGI is not an option" when
nobody has suggested that you use it.

Jon, there's no point in debating Afanasiy about CGI. He's convinced
packages for Python web programming support are inadequate (*all* of them),
is unable to explain what the problem is or what his requirements are, and
doesn't seem to really understand the subject (e.g. thinking FastCGI is
basically the same as regular CGI). Ignoring him is the best course.

--amk

 

[Afanasiy]

Jon, there's no point in debating Afanasiy about CGI. He's convinced
packages for Python web programming support are inadequate (*all* of them),
is unable to explain what the problem is or what his requirements are, and
doesn't seem to really understand the subject (e.g. thinking FastCGI is
basically the same as regular CGI). Ignoring him is the best course.

Read.

 

[Jon Ribbens]

Jon, there's no point in debating Afanasiy about CGI. He's convinced
packages for Python web programming support are inadequate (*all* of them),
is unable to explain what the problem is or what his requirements are, and
doesn't seem to really understand the subject (e.g. thinking FastCGI is
basically the same as regular CGI). Ignoring him is the best course.

Oh, ok, ta ;-)

 

[Daniel Dittmar]

At PyCon, it was generally agreed at the Web programming BoF that the
authors of the various Python frameworks should all implement the same
example application so that potential users can compare the resulting
code. Nothing much has been done on this front since then; we should
get things moving again by figuring out what the example application
should be. The Java Pet Store was suggested, but it was pointed out
that it's a very large application, requiring too much effort for an
author to do in their spare time.

Ian Bicking did something like this with
http://colorstudy.com/docs/shootout.html with a Wiki as the example
application. If there is interest, I would volunteer to copy this into the
Wiki so that it can be more easily extended. I already got Ian's
permissionto do so, but was so far too lazy for it.

Let's think about the requirements for an example application:

* Should be implementable in a few evenings
* Should exercise a reasonable set of features.
* HTML generation (duh!)
* Accepting a form
* Returning non-HTML files (generating a GIF, say)
* Sessions

* access control


I suggest using
http://www.python.org/cgi-bin/moinmoin/WebProgrammingShootOut as the entry
point for the specification of the application.

Daniel

 

[Ian Bicking]

Ian Bicking did something like this with
http://colorstudy.com/docs/shootout.html with a Wiki as the example
application. If there is interest, I would volunteer to copy this into the
Wiki so that it can be more easily extended. I already got Ian's
permissionto do so, but was so far too lazy for it.

It might be preferable to do this in CVS somewhere -- it would handle
the source files better, and most of the commentary can go in comments.

One of the big problems I also had was installation. Many of the
frameworks require non-trivial Apache configuration, which hinders
experimentation. Actually solving this is significant work, though --
but also something which very much deserves solving.

A SourceForge (or equivalent) project might be a good place to start.

Ian

 

[Lothar Scholz]

The Java Pet Store was suggested, but it was pointed out that it's a very
large application, requiring too much effort for an author to do in their
spare time.

But you can see the difference only in larger application. Not
everything scales good when programming in the large is necessary. I
never found that simple benchmarks really helps.

I think I like the store best, because the first three applications are
all text-oriented, and the content manager doesn't do much beyond spitting back

Yes, me too. It should be something that is usefull for comparison in
larger commercial applications. And it should have test generators to
check the system under heavy system load.

 

[Aahz]

That reminds me: there was also a proposal to revive the Web-SIG. Ian,
anything moving on this front? (My offer to host a list still stands.)

I'd suggest putting it on python.org, so that anyone who who looks at
the mailing lists will find it.

 

[Paul Rubin]

The Java Pet Store was suggested, but it was pointed out that it's a very
large application, requiring too much effort for an author to do in their
spare time.

Heh, the point of doing it in Python is it should take much less time
than doing it in Java .

* Should be implementable in a few evenings
Good

* Should exercise a reasonable set of features.
* HTML generation (duh!)
* Accepting a form

Accepting a very long form (i.e. megabytes of POST data)

* Returning non-HTML files (generating a GIF, say)
* Sessions
* Uploading a file
* RDBMS access? (That's more a function of the DB-API module you're
using, but a framework might provide support for pooling
connections or something similar.)

Yes, this should be included, lack of a standard DB API module is a
serious Python deficiency and any DB-using app needs to work around
the deficiency somehow.

* Other suggestions?

Possibilities:
* A Wiki?
* A multiple-user weblog?
* A Slashdot-like discussion board?
* A simple content manager -- upload files, set permissions on them,
control access to them, and download them again.
* A simple browse-a-catalog-and-buy-things store?
* Other suggestions?

Webmail client