The generation process generates all sorts of files containing data about each topic (e.g. the search index). You could try generating to a different location and copy / paste the topic's output file into the previous published location, but the danger is that one of the other files was generated based on the topic's content at that time could have incorrect / invalid data. The generation process doesn't take too long even for some of my large projects. Is this a issue for you?
Thanks Colum. I'm already doing that anyway, since we use Subversion for version control; so I have to generate the output to one (empty) location and then copy all those files to the location under version control, before committing the changes from there to the SVN repository.
No, the length of time isn't a big issue. It's more a matter that sometimes a large commit to SVN will fail after a few minutes, so I have to recommit (sometimes more than once). So what I eventually did today for my "one file changed" situation, was to generate everything to an empty location, as usual, and then copy only the changed file to the area from which to commit the change to the repository.
But thanks for the reply, especially so soon. I'm a fan of your writing, and actually went and browsed your site for a possible answer before posting here.
Just a few questions here.
Is the avoidance of generating the entire project related to the Source Control issue or is it related to a publishing issue? For example, you wish to put the single changed file in place on the serverm but you dislike having to copy them all?
If it's a "copy to server" issue, I might suggest you investigate using the Publish function. The first time you publish, all files are copied from your output folder to the server, But with each subsequent publish action, only the changed files that actually need to be copied are copied.
If it's related to source control, I suppose you could accomplish the same by simply amending your process a bit. Generate to the empty location, but configure publishing so it publishes from the empty location to your repository. Then only the necessary files are changed.
From what you describe, it seems like overwriting one file in your local repository gets around any SVN Commit issues. Is this right? If so, it maybe worth exploring where your original SVN repository is. We use SVN and have some fairly hefty RH projects also. We don't have any of the issues you describe.
A further question if we may. Is the problem happening with more than one project? If only one, can you tell us a little more about it. For example the number of topics or whether it contains large numbers of baggage files, images, etc.
Thanks, Rick. Using SVN means that when files are deployed it is only the changed files that are pushed to the web server. Using Publish isn't an option, SVN is vital for corporate version control. More in my reply to RoboColum, to save duplication here.
Thanks, Colum. Only one project involved; several hundred topics, few baggage files, few images. SVN repository is on a different continent. You're right, I don't get the issues committing output if I commit only a few. (And they're rare enough anyway.) I was just tryng to simplify the process for when there are only a few files (especially just one), and thought that RoboHelp might have a selective Generate option; but I certainly understand why it doesn't, since search wouild be affected even if navgiation wasn't.
Thanks to you and Rick for your help, anyway.