How do I split a very large RH8 project?

[hemtya]

Hi,

I need to split a large RH8 (webhelp) project with about 10,000 files. There is no logical way to divide the contents, the topics are all intertwined. However, there is one folder that contains 3000+ files.

Have anyone split a large project like this? Appreciate any tips, best practices or suggestions.

Thanks, Hemtya

[RoboColum_n_]

Hi Hemtya. I've always tried to avoid getting into this situation but I'd do the following:

• Think carefully how you'd divide the project - maybe by folder is a good way although I'd also divide the 3000 file folder also.
• Create another project to act as a Master into which all the others are merged.
• Copy the project to a new directory, open RH and delete unwanted topics (e.g. by folder).

There is lots of other stuff to consider (e.g. path links) but how you tackle these things depends on your output type. If you tell us that, we'll be able to point you to some other resources.

---

The RoboColum(n)

@robocolumn

Colum McAndrew

[Peter Grainge · Accepted solution]

This explains how I tackled splitting a single project.

http://www.grainge.org/pages/authoring/merging_webhelp/splitting.htm

---

See www.grainge.org for RoboHelp and Authoring tips

@petergrainge

[hemtya]

Hi Andrew,

Thanks for helping. I inherited it, but added about 500 topics not able to figure out how to safely split it. Here are some details.

• The output is webhelp.
  
• There are a total to 10 folders, of which three folders have grown out of control, say Folder X with 3000 files, folder Y with 1000 files and folder Images with over 4000 images called by files in folder X and Y.
• To add to this complexity, some topics in the 'Other7Topics' call the images randomly, in bunches.
  

I do not want to divide it into too many child projects in fear of spending too much time managing/resolving issues between the child projects.

Is it OK to have four child projects: Project-Folder X, Project-Folder Y, Project-Images, Project-Other7Topics?

Peter, thanks. Your article was very helpful. I went over it earlier and created a mock setup to get a hang of the process. However, I need help in figuring out:

• The best logic to divide the topics.
• How many child projects should I have? NOTE: Going further, this guide could grow insanely big based on business requirements that are totally unpredictable.
  

Thanks for all your help,

Hema

[RoboColum_n_]

It is difficult for someone not close to the project in question to suggest the correct split project size. You'll have to do some homework here based on the feedback already offered and your knowledge of the existing project. TBH once you have made the decision to have child projects, whether you have 3 or 300 of them makes little difference. Peter's site has all you need to know for WebHelp set-up. You'd be wise to bookmark it now!

I'm not sure what you mean by "calling the images randomly". Random logic does not bode well in the IT world. Logic rules. I'm sure it does for you also so perhaps you could expand on this.

---

The RoboColum(n)

@robocolumn

Colum McAndrew

[MergeThis]

I can only see one way of splitting the monster, and that's following your current TOC. After all, you will be creating the merged project by adding project identifiers into the master's TOC. Each of the project TOCs then merge to appear as one.

If there is to be any restructuring, you'll have to decide whether to do it in the monster first, or do it as you go. The beauty of merged WebHelp is that the entire help does not need compiling/generating to reflect changes in the output; you're generating and publishing only the active project to its own output folder. The output fo all other child projects remains the same.

I followed Peter's method seven years ago, and it still works as well now.

Good luck,

Leon

[hemtya]

Thanks for the direction so far. My problem is with arriving at best stragtegy to split the project.

The structure if my project is, say, a set of three top layer units, with 100 topics each, and a set of 3 major second layer units, with 3000 topics each. All 6 units are interconnected. So any which way I divide the project, they will have several (external) links to the other child projects.

With this structure, is it even recommended to split the project, meaning is is going to help reduce the complexity at all? If yes, what is the best (less messy) way to do it? I cannot see any logical way to split it.

Any thoughts?

As always, appreciate any help.

Hemtya

[Peter Grainge]

Your original project was 10,000 topics and at that size it is likely you will have performance issues. I am guessing that was your rationale for splitting in the first place. Whatever you do is going to have some work involved as what were topic links now become cross project links.

I cover how I did that on my site but it is a process where you have to think about it carefully and accept that some of the trial and error processes will result in error. If you don't have too many links that will be cross project, it may be easier to find them with something like Xenu and then fix them manually.

As to the split of topics, Leon suggested using your TOC to indicate the logical groupings. Beyond that it is really only you who can work that out based on the content of the topics.

---

See www.grainge.org for RoboHelp and Authoring tips

@petergrainge

[hemtya]

Thanks. I am now trying the split. My question is, in "deleting topics that are not part of the project" phase, I am able to delete only topic by topic. When I try to do a batch delete from the "Topic List" pod, it also deletes the refereneces by default - doesn't give me the choice to retain the references

Is there a way to delete them in batches?

Thanks for all help.

[Peter Grainge]

Not if you want the options.

---

See www.grainge.org for RoboHelp and Authoring tips

@petergrainge

[hemtya]

Hi Peter,Colum, Leon:

Finally, I successfully split my project into two. Many thanks to all your suggestions. Many many many thanks to Peter for his instructions - I followed it like a text book, which in fact it is! I will have to further divide the projects to have them in manageable chunks, which I am sure will be a lot easier than the intital split.

My shortcut to deleting topics in the child projects:

Deleting files and folders through RH was next to impossible. It took about 1 min to delete a topic, and I have some 9000 topics to kill. After 3 days of failed attempts (RH hung up regularly), I figured I can trick RH by deleting the topic/folder references from the root.fjp, etc. I needed the reference to be intact anyways, so it worked. However, the index entries were also retained since RH was oblivious of the deletion. Fortunately, my split logic required to either keep them all or loose them all. So it worked fine too.

Question:

1. Though all external topic references are  resolved in the result, a bunch of links are reported broken in one child project. I did them over twice, but they would budge. Interestingly, the links resolve perfectly within robohelp and in the merged output as well. Any thoughts why?
2. If I do not publish, what are my options for external link validation. All tools seem to help test a published site.

Thanks, Hemtya

[Peter Grainge]

I suggest you try Xenu as that should help with both issues.

It will work with the generated output so the fact that you do not publish is not a problem. I don't have it on this machine to check but from memory you just go to file open and browse to the start page of the generated output. Use the filters to show only broken links.

If you then get Xenu to run one of the reports, you will find the information there is more helpful in finding the broken links. They are there somewhere.

---

See www.grainge.org for RoboHelp and Authoring tips

@petergrainge

[MergeThis]

A few tips on Xenu (other than being put off by their goofy alien):

Options > Basic tab

• Broken links, ordered by links: "This broken link appears in these topics."
• Broken links, ordered by page: "This page (topic) has broken links to these topics."
• Broken local links: "This topic has broken links to internal bookmarks."

You'll have to run some tests to determine whether to choose "by link" or "by page." If one or more writers have made extensive renaming/restructuring, the "by links" option might be the most helpful.

Options > Advanced tab

Exclude MSO files: If file originated as Word file, ignore all MSO identifiers.

Report

The default file type is .mht, which looks a bit weird and which I haven't been able to globally change. I always have to remember to select .htm. There'll also be some false positives, such as to external web sites and files that are only accessible in final builds and not in whatever staging area you might be running Xenu against.

We run three to four software versions simultaneously, and I run reports fairly often; therefore, I keep separate folders/files with the use of this naming scheme:

Folder: yyyy_mm_dd_Vxxx

File: Vxxx Broken links Month day, year.htm

Good luck,

Leon

[hemtya]

Leon, Peter,

Thanks. Xenu works fine. Thanks for the tips on the options. That helped.

I also remembered a tool that I had used in the past, a shareware called HTMLLinkValidator. I am comparing the two, and leaning towards HLV. I will update on the results.

Thanks, Hemtya

[Peter Grainge]

Xenu was suggested as it is free. If you have something else that does the job, go with it.

---

See www.grainge.org for RoboHelp and Authoring tips

@petergrainge