sajolida:
> Emma: How frequently do you see translators mess up with the HTML in
> strings on Weblate?
VERY often
> Can the incorrect HTML be spotted automatically
> somewhere down the line?
Yes, I believe that the simpler and shorter the string, the less likely a translator will make an error.
I also think:
- For me it is hard to read the lists, specially the differences when only one line changes in a long list, i.e.
https://translate.tails.boum.org/translate/tails/license/es/?type=fuzzy
The lists can be made fuzzy adding a space before in each line. We talked about this on this list I think....
- Also sometimes in <pres>:
https://translate.tails.boum.org/translate/tails/finances/es/?type=fuzzy
too big!
I keep answering inline:
>> On 09/09/2018 03:31 PM, sajolida wrote:
>> [...]>
>>> Only some days ago did I realize that this might be a real pain for
>>> translators to translate and maintain. Of course, you've been dealing
>>> with that for years and already know how to do it well, but what about
>>> newer translators and future teams?
>>>
>>> I haven't done much translations myself and I've never used Weblate, so
>>> I'd like to know how much of a relief it would be if we stopped writing
>>> this way and used Markdown instead.
- Weblate has some XMl validation tests, but they don't work with Markdown.
https://translate.tails.boum.org/checks/?project=tails (all fixed currently) so currently we make some mistakes when translating that break the page, like
translating [[link]] by [[somenewwordnotarealpage]] or [[link] without the right amount of [.
>> I have some experience with html, so for me its not that hard to parse
>> those, but i can see how it might lead to confusion and/or mistakes and
>> changing it to pure markdown would definitly make it easier to read. How
>> hard it is also depends on the strings before and after the
>> application-span.
Many translators translate class names, as in <div class='next'> becomes <div class='prochaine'>
>> I think that changing it to pure markdown would definitly be an improvement.
I am not so sure about that, I think it is important to add comments and make it as simple as possible!
>>> - Whether it depends on your translation tools, for example, if PO Edit
>>> is affected in the same way as Weblate. They might be easier to
>>> translate on some tool than other or some tools might detect broken
>>> HTML.
>>
>> I haven't used weblate that much, so i'm not sure if its easier or
>> harder. At least in vim and poedit it doesn't make a difference. Poedit
>> allows to just copy the source string using a Ctrl+b and then modify it
>> and i see that weblate does the same with Ctrl+o- this make it easier,
>> because you can just ignore the html part (you don't have to copy it by
>> hand) and concentrate on the content.
>
> Ok, good to know.
>
> What about strings updates. For example, when you want to unfuzz a
> string I guess it's sometimes better to fix the string that is already
> translated without going back to a copy of the original string in
> English, right?
I think weblate helps a lot on this, because you get presented with the changes on the source github-style, so it makes it easier to decide where to look, see for example:
https://translate.tails.boum.org/translate/tails/known_issues/es/?checksum=8e7c3195c5e4513b#suggestions
(the suggestion also has the diff css)
>>> - Hearing from how do this goes for the less technical of our
>>> translators. Say someone who doesn't otherwise know HTML. How frequent
>>> is it for newer translators to made mistakes on these?
Very frequent
>>> - Whether you think that Markdown instead of HTML would lower the error
>>> rates. Because I still want application names to display in italic :)
I am not so sure about this. I think html gets checked by weblate better.
>> are there even errors with html? i can not recall any...
HA! dont get me started! forgotten </span> tags, etc....
> Your HTML can be invalid and they all kind of bads things can happen.
> There are some services to validate HTML for example:
>
> https://validator.w3.org/
>
>> maybe a bit off topic, but for me the hardest to read are string like:
>> [[!meta title="Other Linux (Red Hat, Fedora, etc.)"]] [[!meta
>> robots="noindex"]] [[!meta stylesheet="bootstrap.min" rel="stylesheet"
>> title=""]] [[!meta stylesheet="inc/stylesheets/assistant"
>> rel="stylesheet" title=""]] [[!meta stylesheet="inc/stylesheets/linux"
>> rel="stylesheet" title=""]]
>
> Excellent point! Let's see if we can fix this.
>
> These are also a pain for me because whenever I want to change one of
> the meta directives, it breaks translations for all the others.
>
