PGDN source browser

On 04/06/2005 22:59:19, Gevik babakhani wrote:

Dear all,

The PostgreSQL Developer Network's Source Browser (beta1) is ready.

If you got the time to check it for a moment, please do not hesitate to send

Firstly, good that you're asking for comments (I assume as an attempt to
work with people), but starting an entirley seperatre "developer network"
seems rather like reinventing the wheel. Why not invest your time in
extending the existing rersources, rather than starting from scratch?

Other than saving me from downloading the source (which I keep around
anyway, mostly for ease of grepping for things the documentation cant be
expected to cover), I see no advantage to using this in it's current
form.

To me, the point of a source browsing system is that it provides
information which a directory structure can't provide. Off the top of my
head:

* CVS head. Without this, it's always going to be irrelevant to
developers who commit to postgresql itself.
* CVS history. If you're targeting this at developers, I'd think this is
important.
* Referencing between function calls and definitions (with ctags,
perhaps. Doxygen-style browsers do this.) To me, this is the single
only advantage of presenting source code in HTML: ease of navigation
by hyperlinks. Apart from that, web pages are pretty inconvenient (to
me, at least).
* Is a tree really appropiate? To compare two files (which is something
I might want to use this for), that'd require a lot of scrolling to
see where I am in the structure.
* diff.

Meanwhile, some aesthetic things which spring to mind:

The syntax hilighting is confusing for non-.c files; quite a few are
parsed incorrectly around comments, and hilighting applied to strange
things in plain-text files.

There is extreneous whitespace in the <pre> at the top by the line
numbers' gutters. There's a off-by-one error in your loop to pad out the
numbers: an extra space appears every power of 10.

The "number of views" is irrelevant, as is the "...Source Browser" title
on each page, which is unneccessary.

Hope that helps,

Firstly, good that you're asking for comments (I assume as an attempt to
work with people), but starting an entirley seperatre "developer network"
seems rather like reinventing the wheel. Why not invest your time in
extending the existing rersources, rather than starting from scratch?

Other than saving me from downloading the source (which I keep around
anyway, mostly for ease of grepping for things the documentation cant be
expected to cover), I see no advantage to using this in it's current
form.

To me, the point of a source browsing system is that it provides
information which a directory structure can't provide. Off the top of my
head:

* CVS head. Without this, it's always going to be irrelevant to
developers who commit to postgresql itself.
* CVS history. If you're targeting this at developers, I'd think this is
important.
* Referencing between function calls and definitions (with ctags,
perhaps. Doxygen-style browsers do this.) To me, this is the single
only advantage of presenting source code in HTML: ease of navigation
by hyperlinks. Apart from that, web pages are pretty inconvenient (to
me, at least).
* Is a tree really appropiate? To compare two files (which is something
I might want to use this for), that'd require a lot of scrolling to
see where I am in the structure.
* diff.

Meanwhile, some aesthetic things which spring to mind:

The syntax hilighting is confusing for non-.c files; quite a few are
parsed incorrectly around comments, and hilighting applied to strange
things in plain-text files.

There is extreneous whitespace in the <pre> at the top by the line
numbers' gutters. There's a off-by-one error in your loop to pad out the
numbers: an extra space appears every power of 10.

The "number of views" is irrelevant, as is the "...Source Browser" title
on each page, which is unneccessary.

Hope that helps,

Regarding to update the techdocs and not reinveting the wheel, I am having a
mailng with Robert Treat of (webmaster pg) in order to see what can be
done.. any help is welcome... :)

Regards,
Gevik.

-----Original Message-----
From: Jonah H. Harris [mailto:jharris@tvi.edu]
Sent: Monday, June 06, 2005 4:50 PM
To: Tom Flavel
Cc: Gevik babakhani; pgsql-hackers@postgresql.org
Subject: Re: [HACKERS] PGDN source browser

Gevik,

You still didn't answer my question as to why you're reinventing the
wheel. Why not just work on updating techdocs instead?

-Jonah

Tom Flavel wrote:

On 04/06/2005 22:59:19, Gevik babakhani wrote:

Dear all,

The PostgreSQL Developer Network's Source Browser (beta1) is ready.

If you got the time to check it for a moment, please do not hesitate to

send

Firstly, good that you're asking for comments (I assume as an attempt to
work with people), but starting an entirley seperatre "developer network"
seems rather like reinventing the wheel. Why not invest your time in
extending the existing rersources, rather than starting from scratch?

Other than saving me from downloading the source (which I keep around
anyway, mostly for ease of grepping for things the documentation cant be
expected to cover), I see no advantage to using this in it's current
form.

To me, the point of a source browsing system is that it provides
information which a directory structure can't provide. Off the top of my
head:

* CVS head. Without this, it's always going to be irrelevant to
developers who commit to postgresql itself.
* CVS history. If you're targeting this at developers, I'd think this is
important.
* Referencing between function calls and definitions (with ctags,
perhaps. Doxygen-style browsers do this.) To me, this is the single
only advantage of presenting source code in HTML: ease of navigation
by hyperlinks. Apart from that, web pages are pretty inconvenient (to
me, at least).
* Is a tree really appropiate? To compare two files (which is something
I might want to use this for), that'd require a lot of scrolling to
see where I am in the structure.
* diff.

Meanwhile, some aesthetic things which spring to mind:

The syntax hilighting is confusing for non-.c files; quite a few are
parsed incorrectly around comments, and hilighting applied to strange
things in plain-text files.

There is extreneous whitespace in the <pre> at the top by the line
numbers' gutters. There's a off-by-one error in your loop to pad out the
numbers: an extra space appears every power of 10.

The "number of views" is irrelevant, as is the "...Source Browser" title
on each page, which is unneccessary.

Hope that helps,