Why do files drop out of WebHelp builds when they exist in the original project?

[shhimwriting]

I am using RH 10 (upgraded early June 2013), Windows 7 and we support IE 9, Chrome, and Firefox.

I build my WebHelp to a WebHelpGenerated folder before checking it into CVS for my development to pull and place in our product.

My project has 1200 files.

I update a topic, save, generate, and lately the build has missing topic pages.

They display in the TOC, but the file is missing altogether and when I check the output View in RH10, it is missing there too, but the file is in the existing project folder. Somewhere along the way, the build is not picking up miscellaneous files.

In 2 of 3 cases, in my project folder, I had the original file and what I assume is some kind of duplicate file, that had a tilde symbol, for example, instructions~.htm. I didn't have permissions to do anything with this file, but after a reboot, I could delete the file, rebuild the Help and everything would be fine. In one case, I didn't have the odd file, but couldn't get the missing file in the build until a reboot.

Any help you can provide would be greatly appreciated.

Thanks

[Captiv8r]

Hi there

Files don't typically just "fall out" of a build. Are you certain that you aren't using some sort of build expression along with Conditional Build Tags that is causing this to happen?

Cheers... Rick

[shhimwriting]

Perhaps "fall out" was the wrong way to describe, but random files are not getting built so when you try to open the topic, you get a "page not found" error, since the htm is missing from the output file.

To answer you question, no, I don't use build tags.

Thanks

[Captiv8r]

Hello again

So this is happening immediately after you build WebHelp before you place it on the server? Or is it happening after you upload to the server?

Cheers... Rick

[shhimwriting]

Hi,

Yes, immediate. Once I discovered the first missing topic, I went back and looked at my Output View and the file was missing there too. On subsequent builds, where I was checking to see if the problem would persist, I knew it to be immediate since I had a specific number of topics that I updated for the day and could double check that they were being processed. This has occurred now on three different days in the past week in my project and once in a co-worker's separate project.

We do not upload to a server, we build locally and check in to a repository ( CVS ) and our developers pull the files and place in our products.

Thanks

[Captiv8r]

Can you use Jing or Screenr to record the process of showing us a walk through of your output settings? That may provide a clue as to what is happening here.

Cheers... Rick

[RuWriter]

Wow! This is the exact problem my tech writing team is having; that's something that two different people reported the same issue today. I am RuWriter.

Captiv8r replied to my post (HTM topics mysteriously missing from compiled WebHelp using RH 10) telling me I shouldn't post the same question twice, but I actually am a different person from shhimwriting. I had done several searches over the past few days to try to find if this issue had been reported before anywhere on the forum, so I'm sorry I missed that someone had posted this issue earlier today.

I'm not sure if it's appropriate to go ahead and comment on my issue here? but since they are so similar and I know it will take everyone extra time to reply in two places, I will add mine here.

We have been seeing every issue that shhimwriting has seen. The problem we are having occurs immediately after building, before we do anything with the output files, and we do not use any Conditional Build Tags. The Output pod in RH does not include the missing topic in the list of ones that have been updated.

We are using the same output settings as usual, so I'm not sure why this would start happening now. I don't think I have a way to record the screen, but I'll try posting images of our settings.

Thanks!

RuWriter

[Captiv8r]

Uh... you and shhimwriting have the same company email address suffix of booksys.com. So both your message as well as shhimwriting's is from the same company. (Never try to fool a forum moderator. LOL)

In the first screen I see the drive letter is E:. Are we to assume this to be a network location? I'm guessing it is. And if so, would this be the same location the end users would be accessing the files from?

If so, the answer here is most likely to change that E: location to be a location where your project lives. Best choice for that would be C:\Projects\SomeFolder. And the WebHelp output would be something like: C:\Projects\SomeFolder\!SSL!\WebHelp

Cheers... Rick

[RuWriter]

Well, that's embarrassing! I apologize to both of you! I didn't know how to see someone's email address and didn't realize that was my supervisor's screen name. Obviously a misunderstanding on my part; I thought one of us was going to contact tech support and one of us would post to the forum. Please excuse me!

We do generate to a virtual E drive, which is a network location, but that is mapped to our C drive. Our developers instructed us to set up our paths this way, and we've been doing it like that for a long time. Neither place is where users pull the Help files. We check them in to CVS, and our software developers then link to them from the product and generate that.

RuWriter

[Peter Grainge]

Rick is (or should be) tucked up in bed right now so let me pick up.

First it is only moderators who can see the email addresses.

The workflow you need is to generate to a local drive and then publish to the E drive. That will get your output to the location that your developers have specified and hopefully without the issue you have encountered.

---

See www.grainge.org for RoboHelp and Authoring tips

@petergrainge

[shhimwriting]

Good morning,

We are generating local, we are using a mapped drive E, which is identical to our local C drives. Our development needs for all writers to build to the E drive using the same path, so all the paths are the same for topics and images regardless of author. We check in our build from the C drive so development can pull from CVS. Since we have done this for a while, even with RH8 and didn't experience this issue, we find it hard to believe this is the root cause.

I am updating new files today.

I will see what happens when I build. I don't have tools to capture the generation process, but I can snag the output and original folders if I find missing topics today.

Thanks for your patience and support.

[Captiv8r]

Hi there

You say:

Since we have done this for a while, even with RH8 and didn't experience this issue, we find it hard to believe this is the root cause.

And yet here you are having an odd problem, no? So what changed?

The version of RoboHelp you say? Yep, ANY new version of any software I've ever seen is apt to have unexpected behavior in some respects. Some choose to call these, "features". Now don't get me wrong here. RoboHelp 10 is *supposed* to work better than its predecessors with network drives.

The fact is, you are using some monkey business with drive letters. And to me, that's suspect. As an initial troubleshooting step, I'd simply generate to the C drive and see if the problem clears. It's a very simple way to test. If it clears, you have your answer. If you still see missing files after you switch to C, then something else is afoot and you would be free to change back to the favored E drive to continue.

Cheers... Rick

[RuWriter]

Thanks for the suggestions, everyone. Shhimwriting and I have done several tests to try to figure out what is going on (we are each the lead author for two different projects). I'll try to summarize what we've tested and what our results have been; sorry if this gets long, but I'll try to give you as much detail as I can so it's clear what we did.

Today, I made edits to several topics, saved them, and generated a build as usual to the E drive. One out of three topics I worked on was missing. I tried generating again (same steps, same drive) to see if that would fix it because simply regenerating has fixed the problem in the past. Today, it did not. Next, I tried generating to the C drive. Again, the same file (only one out of the three I modified) was missing from the build on the C drive. Last, I closed all applications, rebooted my machine, and generated again. This time, we agreed to send the output to the E drive again to figure out if the action of rebooting helped regardless of drive, and sure enough, the missing file was compiled correctly using the E drive.

During this testing of mine, shhimwriting had also made changes to several files in her project, generated, and had one missing file using the E drive. She simply rebooted, generated again, and the file was added back in to the output using the E drive without ever testing the C drive. So the act of rebooting seem to fix, even though she was always using her E drive.

Since the reboot seems to be helping, we did another test to see whether the recent reboot made a difference for an additional build. We both made changes for several topics in our separate Help files.

I tried generated to the C drive first this time (since we weren't sure if the fact that we initially generated to E last time was messing up the C drive test). This time, no topics were missing. So, that was good, but again, since our problem has not been consistent (files missing sometimes but not others), we didn't feel reassured that the problem is "fixed" per se. [In my other post, I mentioned that a few days ago, I had tried making changes and generating to deliberately reproduce the problem and could not - always using the E drive.]

Since shhimwriting had also made additional changes after her last build, she tried generating again. In her case, she built to the E drive as usual, and no files were missing.

Hopefully, I haven't totally confused you, but the bottom line is that we have seen files missing immediately after generating to C and E and files not missing immediately after generating to C and E.

The two of us have noticed some differences in our daily edit/save workflow, but since they don't seem to impact whether we have missing files or not (we each get the same results regardless of those specific differences), I won't complicate matters by sharing those right now.

Our theory at this point is that something about Windows or background processes is causing a glitch while RH 10 is trying to generate. It doesn't help that we have upgraded our entire operating system, RH, and screen capture tool all at the same time. There are so many things going on, it's hard to isolate what the issue is...

Let us know if you have any thoughts.

Thanks for wading through these notes!

RuWriter

[Jeff_Coatsworth]

Frankly, I wouldn’t waste my time on it – best practices dictate working in c:\projects\[project_name]\ - if using source control, check out all the project files to your local machine, work on them, generate help locally to the \!SSL!\ folder, publish it wherever; check the project files back in.

Wham, bam, thank-you ma’am! You’re done…

[RuWriter]

Thanks, Jeff. That may be the best thing to do for general practice, but I think our concern with that would be... would that fix this particular problem? Since we have now tested generating to a local C drive and still saw missing files, we feel like there is something else that's messing up the generation process...

Does everyone agree/assume that generating specifically to a "Project" folder on the C drive (and not just anywhere on the C drive) would make a difference? If so, we have not tested that, but we would have to confer with our developers before making a change like that because we check in our project and generated files using CVS and have to have a matching file structure to the directory there.

Thanks!

[Jeff_Coatsworth]

You’ve got 2 complicating factors at play here – 1 is that you’ve got some funky drive mappings going on, so who knows if that’s messing with your process; 2 is that you’re playing with source control & again, who knows if you’ve got that set up correctly.

All I can say is that I’ve never experienced topics missing from my projects’ help build unless I specifically excluded them with CBTs.

[Amebr]

A left field suggestion. Perhaps see if there has been a recent change to your antivirus - say different scanning time, or scanning both incoming and outgoing files. I've seen cases where the AV is scanning a file and locking it for the duration - I don't think I've had the specific issues you see, but it can produce weird behaviour.

Amber

[Peter Grainge]

Correct me if I am wrong but it seems the workflow is you all check stuff out and you all generate to the E drive. Surely the procedure should be that you all check out as required and then check back in, then elect one of you to PUBLISH to the E drive.

---

See www.grainge.org for RoboHelp and Authoring tips

@petergrainge

[RuWriter]

Amber - thanks for the suggestion. This might be something to look into. It looks like there is a way to exclude certain files/folders/processes from being scanned by our anti-virus software, so it may help to do that for our RH work. My supervisor is out of the office today, so I'll have to discuss this with her tomorrow. So far, rebooting before we build (using our regular procedure) is fixing the problem, so we think the issue is related to something about our operating system, in which case antivirus software could be the culprit.

Peter - Two out of three of us generate because there are two completely separate projects. Each project has a lead author, so shhimwriting is the only one who generates her main project, and I am the only one who generates my main project, even though we may have multiple people working in them at times. All of us can potentially "check in" project files (as a backup and means for us to get the latest someone else has worked on), but only two of us "check in" built files for the two separate projects (so development can pull them into the products when they build the software). And again, as far as our structure, we use E for all project and generated files so no matter who is working, the embedded file paths that point to topics, images, etc. match (to ensure that links always know where to pull something from); then, we check in to CVS using C because CVS options are only available from the local hard drive. Sorry if this still isn't clear, but development doesn't "need" the published version to be on E. They only set up our structure that way so that the RH project would always know where to look for and find files. To get the files to development, we check in files from the C drive, and development pulls it from CVS, not from any location on our local machines (C) or company network (E).

Thanks,

RuWriter