Re: [Webmacro-devel] Recent changes [Was Re: [WebMacro-cvs] webmacro/test/unit/org/webmacro TestContext.java, 1.3, 1.4]

[Endre S. <En...@st...>]

Tim Pizey wrote:
> Hi Endre and all
> 
> On Sunday 16 March 2008 17:28, Endre Stølsvik wrote:
>> Tim P wrote:
>>> Update of /cvsroot/webmacro/webmacro/test/unit/org/webmacro
>>> In directory
>>> sc8-pr-cvs9.sourceforge.net:/tmp/cvs-serv15870/test/unit/org/webmacro
>>>
>>> Modified Files:
>>> 	TestContext.java
>>> Log Message:
>>> Eclipse warning: unused variable
> Hmm, I am not getting any CVS messages at the moment, 
> maybe spamassassin is eating them. 
> 
>> It is cool that you fix up WM..
>>
>> Two questions:
>>
>> 1) Have the license change that has been acknowledged by original author
>> been effectuated? -> BSD w/o advertising clause, wasn't it?
> 
> Eugh, licensing is my least favourite topic. 

OK?

Hope for your sake that you don't sell any software then.

> As far as I am concerned anyone can do what they want with my code.
> My understanding is that we are dual Apache/GPL.

Is this mentioned in the headers of the files, and in a README?

Because NO-ONE can really use it properly without this taken properly 
care of. If you intend to use this as your own code, that's of course 
OK. But I won't touch GPL code with a five-fucking-million feet pole 
when it comes to anything remotely commercial.

I do remember VIVIDLY that this was taken care of, and that .. 
whats-his-name ... gave full permission to change the license, as the 
company that now "owns" this stuff is no more, and that he himself 
didn't have any interest in anymore other than that he hoped other could 
ise it. I believe the consensus went with BSD, I believe, WITHOUT the 
advertising clause. BSD is GPL compatible. Apache claims that Apache 2 
is compatible, while the Gnus aren't that sure.

This is in the email archives.

This MUST be taken care of if it is hoped that WebMacro shall again be 
usable to others than the five people on this list.

> 
>> 2) The bug fixes and huge optimizations (as for example removing
>> allocations of huge buffers that was never used) that this
>> dude-out-of-nowhere came up with after profiling WM rather extensively,
>> been stacked into the code?
> 
> I do not know.

Well, since you have the code "active" in your editor, and obviously 
have access to cvs, which I don't, could you care to check it up? IIRC, 
there were _significant_ boosts of the stuff he did. In particular, 
there was some megabytes-char arrays that was just constructed and never 
actually used - in the FastWriter stuff. This had something to do with 
the pooling mechanism that .. isn't (?) any longer used..

> 
>> Thanks,
>> Endre.
> 
> Although I have been involved with WebMacro since the start, it is only 
> recently that I have gained the skills and the tools to actually understand 
> and modify it.  (Hooray for Eclipse/Maven).
> 
> These tools do however find little mistakes and harmless omissions, 
> so all I am doing is aiming to make WebMacro 'clean' in the sense of 
> having no warnings in Eclipse. 
> 
> I have some other, longer term goals:
> 
> 1. When next version of javacc eventually comes out to use that, 
> as it produces fewer warnigs, and a documented version of the grammar.
> I am about to commit a regeneration from the javacc snapshot as it already 
> creates fewer warnings.

What is the big benefit of this?

> 
> 3. Introduce a plugable module which enables the style of template 
> writing which was available in WM 0.98 where white space was eaten in a 
> way which left readable HTML

It is possible to insert spaces and such now also - it is just a PITA. 
However, it was certainly a PITA before also - there is apparently just 
no golden way to do this.

> 
> I have started to scratch my Maven itch by introducing a POM and 
> publishing a project site at http://webmacro.sourceforge.net/ 
> (this url was unused).
> Of particular value/interest is the Cobertura report at 
> http://webmacro.sourceforge.net/cobertura/index.html
> 
> If people are happy then maybe I could become an Admin and 
> make http://webmacro.sourceforge.net/  the homepage until 
> webmacro.org resurfaces.

I'm happy that ANYTHING happens with WebMacro - I find it quite sad that 
it has died.

But if it is hoped that anyone shall use it, the licensing issue MUST be 
taken care of.

> 
> This weekend I tested WM against Velocity for speed.
> The tests of Melati are now nearing completion, so I 
> decided to time them and then change to Velocity and time it. 
> The result, which perhaps should not suprise, is that the difference
> between WM and Velocity are not discernable, ie the variations in running time 
> within WM and Velocity are far greater then between them. 
> Melati tests with WebMacro: 1 min 5/9/11 seconds Velocity 1 Min 4/9/22 
> I ran the tests with internet access disabled but still got 11 second 
> difference in performance within Velocity runs. 

This is actually not the result that the company I worked for got 
testing the same thing.

Webmacro FAR outran Velocity. In pretty much every aspect. Also WebMacro 
has some features that makes certain jobs that was tested against 
extremely much faster. Yeah, wonder if it wasn't something like this: 
WebMacro "streams" the results into a buffer. OTOH, Velocity basically 
does a string-concatenation, resulting in bucketloads of trash. Velo 
doesn't have the Macro class (is it?), so that you cannot make your 
object implement it and thus output into the final stream, you HAVE to 
make it into a string, which then is added to the result (Velo does a 
.toString() on everything, concatenating the result, like a huge 
StringBuffer)

Some changes was suggested to the Velocity list due to this realization, 
so there might have been changes since then (about 1-2 years ago).

Endre.