Hi Norman,

>> - The third way is not very subtle, and frowned upon by CVS gurus, but very 
>> robust (in our case!): You have two working copies like above. You work in 
>> the first one (the trunk). When you're going to publish, you bluntly copy 
>> the file(s) in question over their counterparts in the B_Release working 
>> copy, and commit there (of course you also commit to HEAD). The advantage of 
>> this method is that now you are absolutely sure that the revision in 
>> B_Release is identical to the one in HEAD. You can't make subtle mistakes 
>> while merging, because there's no merging and tagging involved. You just 
>> bulldozer your way into the B_Release branch and drop your stuff there.

> I like blunt and I don't care what the cvs gurus think! ;-) This sounds
> by far the method most likely to cause the least errors! I think I'll be
> sticking with this option.

So am I! Of course, being the snob I am, I started out with the first option, 
but the yokel soon got the upper hand ;-)

With 1 and 2 there's just too much that you mustn't forget, lest you'll find 
yourself in a terrible mess.

> Now, how do I check out the b_release branch please?

On the command line: cvs co -r B_Release

GUI clients usually have a drop-down or text field where you can fill in the 
branch name.

> > Oh, and in all cases: don't forget to build and upload the HTML and PDF :-)
> What is the process these days? I had just got used to SCP'ing
> everything to the right place and "they" went and changed the web site.
> Last I heard was that we didn't yet have a process that was documented.

You have to make a SSH connection to 78.47.172.253, account www-data.

PDFs go into /var/www/firebirdsql.org/staticsite/pdfmanual
HTML into /var/www/firebirdsql.org/staticsite/manual

Both are symlinks. In reality, the PDF dir is 
/var/www/firebirdsql.org/docroot/file/documentation/reference_manuals/user_manuals,
 and the HTML dir 
/var/www/firebirdsql.org/docroot/file/documentation/reference_manuals/user_manuals/html

I take it you don't have access yet? Please send your public key to Sergey 
Mereutsa; I've already asked him to add it to the www-data account when it 
arrives.

> > The advantage of all this: The B_Release branch always contains the latest 
> > publishable revision of every document. So if ever we have to re-upload any 
> > manuals, or even the entire set, all we have to do is build them from 
> > B_Release and we're fine. At the same time, we can commit as often as we 
> > want without fear of breaking any builds.

> Hmmm. I was always of the opinion that you "never broke the build" - so
> half updated documents were "ok" as long as they built. Things that
> don't build are not acceptable. I only ever commit fully buildable
> documents.

So did I. But during the transition, Dmitry rebuilt the entire docset, 
including a number of documents that were "in limbo", so to speak, with 
comments in the text, and symbols like ???? and ## Xxx where there ought to be 
version numbers or dates. That prompted the discussion which resulted in the 
decision to create a release branch. Everything committed to that branch 
(B_Release) must be *publishable* as well as buildable.

> PS. If you let me have the full process for doing options 1 and 2 above
> (tag/merge/update/whatever) and the process for uploading html and pdf
> files to the new website, I'll happily test and then update the Doc
> Writers' manual(s) as appropriate.

Let's forget 1 and 2, unless someone insists on using them.


Cheers,
Paul Vinkenoog

------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security
threats, fraudulent activity and more. Splunk takes this data and makes
sense of it. Business sense. IT sense. Common sense.
http://p.sf.net/sfu/splunk-d2d-oct
_______________________________________________
Firebird-docs mailing list
Firebird-docs@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/firebird-docs

Reply via email to